<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel>
<title>Maxwell Newman</title><link>https://maxwellnewman.com</link><description>Essays by Maxwell Newman.</description>
<item><title>AI Company Guide: Bottom Up, Then Top Down</title><link>https://maxwellnewman.com/ai-company-guide-bottom-up-then-top-down/</link><guid>https://maxwellnewman.com/ai-company-guide-bottom-up-then-top-down/</guid><pubDate>Tue, 13 Oct 2026 00:00:00 GMT</pubDate><description>&lt;h2&gt;Eight months later&lt;/h2&gt;
&lt;p&gt;In February I wrote &lt;a href=&quot;/ai-coding-guide-retirement-is-here/&quot;&gt;AI Coding Guide: Retirement is here&lt;/a&gt;. At the bottom I left a Notes section with what I thought the next steps would be to make it all even more automatic. The last one said:&lt;/p&gt;
&lt;blockquote&gt;
&lt;p&gt;Following that, we start to go truly exponential. We automate ideas. This is where projects do not even need to be thought of before they are fully implemented. [...] This is what I am betting myself and my career on in the next decade.&lt;/p&gt;
&lt;/blockquote&gt;
&lt;p&gt;&amp;quot;In the next decade&amp;quot;... I genuinely thought this was years away.&lt;/p&gt;
&lt;p&gt;It&amp;#39;s October. Let&amp;#39;s check the list:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Done:&lt;/strong&gt; Automate the PM → Claude connection (that&amp;#39;s SSP, more below)&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Done (mostly):&lt;/strong&gt; Automate the code audit (every piece gets reviewed and tested before it merges, and my 1–3 day audit is now a 1-hour review)&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Done:&lt;/strong&gt; Black-box, end-to-end testing to make sure what was built actually works (for my projects, browser and all — still the biggest mountain)&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;In progress:&lt;/strong&gt; Automate the PRD (idea → strategy → product docs — this is what I&amp;#39;m building now)&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Not yet:&lt;/strong&gt; Automate ideas (the idea still comes from me, for now)&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;The first three are all one thing: SuperSuperPowers (SSP), the layer I built on top of Superpowers. More on that below.&lt;/p&gt;
&lt;p&gt;And that created a new problem: I finish projects faster than I can come up with them. I am spending so much of my week just sitting on my butt waiting for inspiration. Which, if you read the &lt;a href=&quot;/ai-coding-guide-retirement-is-here/&quot;&gt;February post&lt;/a&gt;, is exactly the bottleneck I said was coming. I just didn&amp;#39;t think it would come for me this fast. So I&amp;#39;m building the next layer, Goalset, to take me from an idea to the strategy and the product docs.&lt;/p&gt;
&lt;p&gt;So this post is two things. First, a breakdown of every layer that got me here, bottom up. Second, what I&amp;#39;m building now, which goes the other way (top down). I&amp;#39;m writing it down partly so future me (and you, future reader) can see exactly where this started.&lt;/p&gt;
&lt;h2&gt;Part 0: The zoom out&lt;/h2&gt;
&lt;p&gt;Every layer here exists because the layer below it hit a wall. And the wall was always me. Each time the fix was the same: write down the thing I kept doing by hand, then hand it to the machine.&lt;/p&gt;
&lt;figure class=&quot;diagram&quot;&gt;&lt;svg viewBox=&quot;0 0 760 448&quot; role=&quot;img&quot; aria-label=&quot;Goalset is the new layer on top: it decides what to build and why&quot; font-size=&quot;13&quot;&gt;&lt;defs&gt;&lt;marker id=&quot;ladder-arrow&quot; viewBox=&quot;0 0 10 10&quot; refX=&quot;9&quot; refY=&quot;5&quot; markerWidth=&quot;6&quot; markerHeight=&quot;6&quot; orient=&quot;auto-start-reverse&quot;&gt;&lt;path d=&quot;M0 0L10 5L0 10z&quot; fill=&quot;var(--cds-chart-axis)&quot;&gt;&lt;/path&gt;&lt;/marker&gt;&lt;marker id=&quot;ladder-down&quot; viewBox=&quot;0 0 10 10&quot; refX=&quot;9&quot; refY=&quot;5&quot; markerWidth=&quot;6&quot; markerHeight=&quot;6&quot; orient=&quot;auto-start-reverse&quot;&gt;&lt;path d=&quot;M0 0L10 5L0 10z&quot; fill=&quot;var(--cds-chart-categorical-1)&quot;&gt;&lt;/path&gt;&lt;/marker&gt;&lt;/defs&gt;&lt;text x=&quot;24&quot; y=&quot;30&quot; font-size=&quot;15&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset is the new layer on top: it decides what to build and why&lt;/text&gt;&lt;g&gt;&lt;rect x=&quot;24&quot; y=&quot;56&quot; width=&quot;548&quot; height=&quot;80&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;2&quot;&gt;&lt;/rect&gt;&lt;rect x=&quot;472&quot; y=&quot;66&quot; width=&quot;84&quot; height=&quot;20&quot; rx=&quot;10&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.25&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;514&quot; y=&quot;80&quot; text-anchor=&quot;middle&quot; font-size=&quot;11&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;In progress&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;81&quot; font-size=&quot;14&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;1. Goalset&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;104&quot; font-size=&quot;12&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;Goal selection, strategy, research, feasibility, subgoals,&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;122&quot; font-size=&quot;12&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;product requirements, project prioritization, evidence, and replanning.&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;24&quot; y=&quot;152&quot; width=&quot;548&quot; height=&quot;80&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;rect x=&quot;488&quot; y=&quot;162&quot; width=&quot;68&quot; height=&quot;20&quot; rx=&quot;10&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;522&quot; y=&quot;176&quot; text-anchor=&quot;middle&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Existing&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;177&quot; font-size=&quot;14&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;2. SuperSuperPowers (SSP)&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;200&quot; font-size=&quot;12&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;Project planning, orchestration, agent management, testing,&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;218&quot; font-size=&quot;12&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;and completion verification.&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;24&quot; y=&quot;248&quot; width=&quot;548&quot; height=&quot;80&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;rect x=&quot;488&quot; y=&quot;258&quot; width=&quot;68&quot; height=&quot;20&quot; rx=&quot;10&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;522&quot; y=&quot;272&quot; text-anchor=&quot;middle&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Existing&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;273&quot; font-size=&quot;14&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;3. Superpowers&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;296&quot; font-size=&quot;12&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;Structured software development and implementation workflows.&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;24&quot; y=&quot;344&quot; width=&quot;548&quot; height=&quot;80&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;rect x=&quot;488&quot; y=&quot;354&quot; width=&quot;68&quot; height=&quot;20&quot; rx=&quot;10&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;522&quot; y=&quot;368&quot; text-anchor=&quot;middle&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Existing&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;369&quot; font-size=&quot;14&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;4. Models + Claude Code (and other tools)&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;392&quot; font-size=&quot;12&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;Raw intelligence, plus code generation, editing, tool execution, and implementation.&lt;/text&gt;&lt;/g&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;path d=&quot;M592 424V156&quot; marker-end=&quot;url(#ladder-arrow)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;text x=&quot;604&quot; y=&quot;378&quot; font-size=&quot;11.5&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;bottom up&lt;/text&gt;&lt;text x=&quot;604&quot; y=&quot;396&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;we know what we&lt;/text&gt;&lt;text x=&quot;604&quot; y=&quot;412&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;want to build&lt;/text&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot;&gt;&lt;path d=&quot;M720 60V192&quot; marker-end=&quot;url(#ladder-down)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;text x=&quot;708&quot; y=&quot;92&quot; text-anchor=&quot;end&quot; font-size=&quot;11.5&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;top down&lt;/text&gt;&lt;text x=&quot;708&quot; y=&quot;108&quot; text-anchor=&quot;end&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;we need to figure&lt;/text&gt;&lt;text x=&quot;708&quot; y=&quot;124&quot; text-anchor=&quot;end&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;out what we want&lt;/text&gt;&lt;text x=&quot;708&quot; y=&quot;140&quot; text-anchor=&quot;end&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;to build&lt;/text&gt;&lt;/svg&gt;&lt;figcaption&gt;The ladder · four layers, two directions&lt;/figcaption&gt;&lt;/figure&gt;

&lt;p&gt;Read it from the bottom up. Each rung shrinks what the human does.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;The models.&lt;/strong&gt; Raw intelligence. A model can write a function, explain a bug, draft a plan. But it can&amp;#39;t hold an entire project in its head, and it doesn&amp;#39;t do anything until you ask it to. Where I stepped in: everywhere.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Claude Code.&lt;/strong&gt; The runtime. It wraps the model in a loop that can read files, run commands, look things up and fix its own small mistakes. I&amp;#39;ve been using it since day 0 (and before that, my own ghetto MCP version of it). For a solid 6+ months it was the only thing on the market that sped engineering up another 5-10x. Where I stepped in: every 10-30 minutes, 8h a day, to tell it what&amp;#39;s next and catch what it broke.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Superpowers.&lt;/strong&gt; A workflow on top of Claude Code that front-loads all the questioning, research and planning before a single line of code is written. Humans show up at the beginning and at the end. That&amp;#39;s it. Where I stepped in: answering its questions (or, let&amp;#39;s be honest, pasting them to ChatGPT) and auditing the result.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;SuperSuperPowers (SSP).&lt;/strong&gt; My creation. It runs a bunch of Superpowers instances on one giant project, one after another and at the same time, and answers most of their questions using a written record of how I&amp;#39;ve answered them before. Where I step in: one hour up front with a mock of the finished thing, one hour of review at the end.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Goalset.&lt;/strong&gt; The turn. Everything below it builds something you already know you want. Goalset takes an idea and figures out the strategy and what to build.&lt;/p&gt;
&lt;p&gt;Zooming out, here&amp;#39;s what that looked like for me, one layer at a time:&lt;/p&gt;
&lt;figure class=&quot;diagram&quot;&gt;&lt;svg viewBox=&quot;0 0 760 560&quot; role=&quot;img&quot; aria-label=&quot;Each layer took work off my plate&quot; font-size=&quot;11.5&quot;&gt;&lt;defs&gt;&lt;marker id=&quot;era-arrow&quot; viewBox=&quot;0 0 10 10&quot; refX=&quot;9&quot; refY=&quot;5&quot; markerWidth=&quot;6&quot; markerHeight=&quot;6&quot; orient=&quot;auto-start-reverse&quot;&gt;&lt;path d=&quot;M0 0L10 5L0 10z&quot; fill=&quot;var(--cds-chart-axis)&quot;&gt;&lt;/path&gt;&lt;/marker&gt;&lt;/defs&gt;&lt;text x=&quot;24&quot; y=&quot;30&quot; font-size=&quot;15&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Each layer took work off my plate&lt;/text&gt;&lt;text x=&quot;24&quot; y=&quot;50&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;What I did by hand, as each layer arrived · dashed = in progress&lt;/text&gt;&lt;g&gt;&lt;text x=&quot;24&quot; y=&quot;92&quot; font-size=&quot;13.5&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Claude Code&lt;/text&gt;&lt;rect x=&quot;24&quot; y=&quot;108&quot; width=&quot;166&quot; height=&quot;78&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-5)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-5)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;38&quot; y=&quot;132&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Claude Code&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;152&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Writes the code,&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;169&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;one task at a time&lt;/text&gt;&lt;rect x=&quot;24&quot; y=&quot;194&quot; width=&quot;166&quot; height=&quot;294&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;38&quot; y=&quot;218&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Me&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;238&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;8h a day of check-ins&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;258&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Come up with the idea&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;275&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Write the PRD&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;292&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Break it into parts&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;309&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Prompt each part&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;326&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review every 10-30m&lt;/text&gt;&lt;text x=&quot;38&quot; y=&quot;343&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Debug debug debug&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;text x=&quot;206&quot; y=&quot;92&quot; font-size=&quot;13.5&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;+ Superpowers&lt;/text&gt;&lt;rect x=&quot;206&quot; y=&quot;108&quot; width=&quot;166&quot; height=&quot;95&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-4)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-4)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;220&quot; y=&quot;132&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Superpowers&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;152&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Asks questions,&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;169&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;writes the plan,&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;186&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;builds with tests&lt;/text&gt;&lt;rect x=&quot;206&quot; y=&quot;211&quot; width=&quot;166&quot; height=&quot;61&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-5)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-5)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;220&quot; y=&quot;235&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Claude Code&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;255&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Writes the code&lt;/text&gt;&lt;rect x=&quot;206&quot; y=&quot;280&quot; width=&quot;166&quot; height=&quot;208&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;220&quot; y=&quot;304&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Me&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;324&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;A few 2-3h sessions a week&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;344&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Come up with the idea&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;361&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Write the PRD&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;378&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Relay Qs to ChatGPT&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;395&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Kick off each plan&lt;/text&gt;&lt;text x=&quot;220&quot; y=&quot;412&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Audit code (1-3 days)&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;text x=&quot;388&quot; y=&quot;92&quot; font-size=&quot;13.5&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;+ SSP (today)&lt;/text&gt;&lt;rect x=&quot;388&quot; y=&quot;108&quot; width=&quot;166&quot; height=&quot;95&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;402&quot; y=&quot;132&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;SSP&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;152&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Plans the project,&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;169&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;runs Superpowers&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;186&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;in parallel&lt;/text&gt;&lt;rect x=&quot;388&quot; y=&quot;211&quot; width=&quot;166&quot; height=&quot;61&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-4)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-4)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;402&quot; y=&quot;235&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Superpowers&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;255&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Builds subprojects&lt;/text&gt;&lt;rect x=&quot;388&quot; y=&quot;280&quot; width=&quot;166&quot; height=&quot;61&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-5)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-5)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;402&quot; y=&quot;304&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Claude Code&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;324&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Writes the code&lt;/text&gt;&lt;rect x=&quot;388&quot; y=&quot;349&quot; width=&quot;166&quot; height=&quot;139&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;402&quot; y=&quot;373&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Me&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;393&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;2h, twice a week&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;413&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Come up with the idea&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;430&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review the mock (1h)&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;447&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review result (1h)&lt;/text&gt;&lt;text x=&quot;402&quot; y=&quot;464&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Add new rules&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;text x=&quot;570&quot; y=&quot;92&quot; font-size=&quot;13.5&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;+ Goalset (in progress)&lt;/text&gt;&lt;rect x=&quot;570&quot; y=&quot;108&quot; width=&quot;166&quot; height=&quot;78&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;5 4&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;584&quot; y=&quot;132&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset&lt;/text&gt;&lt;text x=&quot;584&quot; y=&quot;152&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Subgoals, strategies,&lt;/text&gt;&lt;text x=&quot;584&quot; y=&quot;169&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;research, projects&lt;/text&gt;&lt;rect x=&quot;570&quot; y=&quot;194&quot; width=&quot;166&quot; height=&quot;61&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;584&quot; y=&quot;218&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;SSP&lt;/text&gt;&lt;text x=&quot;584&quot; y=&quot;238&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Plans each project&lt;/text&gt;&lt;rect x=&quot;570&quot; y=&quot;263&quot; width=&quot;166&quot; height=&quot;61&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-4)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-4)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;584&quot; y=&quot;287&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Superpowers&lt;/text&gt;&lt;text x=&quot;584&quot; y=&quot;307&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Builds subprojects&lt;/text&gt;&lt;rect x=&quot;570&quot; y=&quot;332&quot; width=&quot;166&quot; height=&quot;61&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-5)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-5)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;584&quot; y=&quot;356&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Claude Code&lt;/text&gt;&lt;text x=&quot;584&quot; y=&quot;376&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Writes the code&lt;/text&gt;&lt;rect x=&quot;570&quot; y=&quot;401&quot; width=&quot;166&quot; height=&quot;87&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;5 4&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;584&quot; y=&quot;425&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Me&lt;/text&gt;&lt;text x=&quot;653&quot; y=&quot;467&quot; text-anchor=&quot;middle&quot; font-size=&quot;22&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;relax?&lt;/text&gt;&lt;/g&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;path d=&quot;M24 520H734&quot; marker-end=&quot;url(#era-arrow)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;text x=&quot;734&quot; y=&quot;542&quot; text-anchor=&quot;end&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;over time&lt;/text&gt;&lt;/svg&gt;&lt;figcaption&gt;What I did by hand, layer by layer&lt;/figcaption&gt;&lt;/figure&gt;

&lt;h2&gt;Part 1: Models &amp;amp; Claude Code / Codex&lt;/h2&gt;
&lt;p&gt;The bottom of the ladder is the model. Raw intelligence. It can write a function, explain a bug, draft a plan, and honestly do most of it better than me at this point. But on its own it has no hands. It answers one prompt at a time and forgets everything the second you close the window. Two years ago that meant copy-pasting code out of ChatGPT into my editor and praying it compiled.&lt;/p&gt;
&lt;p&gt;Claude Code (and later Codex) gave it hands. It&amp;#39;s a loop wrapped around the model: read the files, run the command, look at the error, fix it, run it again. It does lookups, it runs the tests, it solves the small stuff on its own. That&amp;#39;s what made it the only product on the market for a solid 6+ months that sped engineering up another 5-10x. The model stopped being something I talked to and became something that actually did the work.&lt;/p&gt;
&lt;p&gt;The unit of work here is a &lt;strong&gt;task&lt;/strong&gt;. And a task is about as big as it can handle. Hand it anything bigger and it starts losing the plot: it forgets a decision it made an hour ago, misses how two pieces connect, and the mistakes cascade.&lt;/p&gt;
&lt;p&gt;So the human ends up being the glue. In &lt;a href=&quot;/ai-coding-guide-retirement-is-here/&quot;&gt;February&lt;/a&gt; I described the standard AI dev process like this:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;Come up with an idea&lt;/li&gt;
&lt;li&gt;Plan it out (in my head, or a PRD)&lt;/li&gt;
&lt;li&gt;Break it into parts&lt;/li&gt;
&lt;li&gt;Ask AI to do the parts one by one (and pray it doesn&amp;#39;t screw it up)&lt;/li&gt;
&lt;li&gt;Review, go down the list&lt;/li&gt;
&lt;li&gt;Repeat 3-5 every 10-30 minutes for 8h a day&lt;/li&gt;
&lt;li&gt;Debug debug debug&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;Every step except 4 is me. The AI was fast. I was the bottleneck.&lt;/p&gt;
&lt;h2&gt;Part 2: Superpowers (move the human to the edges)&lt;/h2&gt;
&lt;p&gt;&lt;a href=&quot;https://github.com/obra/superpowers&quot;&gt;Superpowers&lt;/a&gt; is an open-source set of skills for Claude Code, and if you haven&amp;#39;t tried it yet, stop reading and go install it. Seriously.&lt;/p&gt;
&lt;p&gt;The big idea: most AI coding mistakes are actually planning mistakes. The AI guessed at something it should have asked about, and that guess cascaded into a ton of deeper issues down the line. So Superpowers forces the asking to happen first. Every project goes through the same steps:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Brainstorm.&lt;/strong&gt; Claude asks you questions, one at a time, until it actually understands what you want. It proposes 2-3 approaches and writes a design doc for you to approve.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Write the plan.&lt;/strong&gt; The design becomes an implementation plan of small, concrete tasks. Each one says which files, which test, and what should happen.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Execute.&lt;/strong&gt; Fresh subagents take the tasks one by one. Each writes a failing test first, makes it pass, and gets reviewed before the next task starts.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Verify.&lt;/strong&gt; Nothing is &amp;quot;done&amp;quot; until the tests have actually been run and someone (something?) has read the output.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;The result is that all of the human judgment gets moved to two places: the questions at the beginning and the review at the end. The middle just runs. For hours.&lt;/p&gt;
&lt;h3&gt;How I actually used it&lt;/h3&gt;
&lt;p&gt;If you read the &lt;a href=&quot;/ai-coding-guide-retirement-is-here/&quot;&gt;February post&lt;/a&gt;, you know my workflow. I&amp;#39;d hammer out the product with ChatGPT, have it write a PRD, then open a new ChatGPT window and make &lt;em&gt;it&lt;/em&gt; the project manager. Claude would ask me a question during the brainstorm and I wouldn&amp;#39;t answer it. I&amp;#39;d paste it to my ChatGPT project manager and paste the answer back.&lt;/p&gt;
&lt;p&gt;It worked. It took my 8h days of 10-30m check-ins down to a handful of 2-3h sessions a week. But look at what I was actually doing in that loop: I was a clipboard. A very expensive, very slow clipboard. Copying questions between two AIs, and then spending 1-3 days auditing at the end, catching the same kinds of mistakes over and over.&lt;/p&gt;
&lt;p&gt;And that was the opening. If the mistakes are predictable, and the answers are predictable, why am I the one sitting there?&lt;/p&gt;
&lt;h2&gt;Part 3: SuperSuperPowers (SSP)&lt;/h2&gt;
&lt;p&gt;Superpowers clarifies one task before it builds. SSP clarifies an entire &lt;em&gt;project&lt;/em&gt; before it launches a bunch of tasks. It&amp;#39;s basically a conductor: it plans and coordinates, and Superpowers does the building.&lt;/p&gt;
&lt;p&gt;It comes down to three ideas.&lt;/p&gt;
&lt;h3&gt;1. Write down how you&amp;#39;d answer&lt;/h3&gt;
&lt;p&gt;This is the core of it. After a few months of running Superpowers, I noticed I was answering the same questions and fixing the same mistakes. Every. Single. Time.&lt;/p&gt;
&lt;p&gt;So I started writing down my judgment so it could be applied without me. Almost every rule exists because something went wrong. And the best ones came from problems that didn&amp;#39;t even exist until I had a bunch of agents working at the same time.&lt;/p&gt;
&lt;p&gt;These days, I&amp;#39;ve tested this enough times that I let it go from start → finish without any of my oversight.&lt;/p&gt;
&lt;p&gt;It didn&amp;#39;t start that way. On a new project, SSP starts out with me monitoring everything. Every time it screws something up, that becomes a new rule. And every time it gets something right enough times, I give it a little more rope. Trust gets earned one rule at a time.&lt;/p&gt;
&lt;h3&gt;2. Mock first&lt;/h3&gt;
&lt;p&gt;Before anything gets built, I look at a mock of the finished product. What it looks like, what it does. And I change it as much as I want. This is the cheapest place to be wrong. Changing a mock costs minutes. Changing the real thing costs a rebuild (and a day or two of credits at the minimum, a week at worst).&lt;/p&gt;
&lt;h3&gt;3. One meeting, then it builds&lt;/h3&gt;
&lt;p&gt;SSP does all of the decision-making up front. I read the plan, approve it in one sitting, and it builds for hours without me.&lt;/p&gt;
&lt;h3&gt;The proof: I barely work anymore&lt;/h3&gt;
&lt;p&gt;I&amp;#39;m a founder who was doing the engineering myself to keep up with demand. That meant 12h days, 5 days a week. 60 hours at least. It was the first thing I did in the morning and the last thing I checked before bed.&lt;/p&gt;
&lt;p&gt;Now I work 2 hours, twice a week. 4 hours. And I&amp;#39;m getting more done than I did at 60.&lt;/p&gt;
&lt;p&gt;I am not joking.&lt;/p&gt;
&lt;p&gt;Goalset (the reason I&amp;#39;m writing this post) is one example. SSP built the first version in one day with 350 commits, and my part was one hour of ideation and one hour of review. In &lt;a href=&quot;/ai-coding-guide-retirement-is-here/&quot;&gt;February&lt;/a&gt; the audit at the end took me 1-3 days. Now my hour is spent asking:&lt;/p&gt;
&lt;p class=&quot;callout&quot;&gt;&quot;did it build the &lt;strong&gt;&lt;u&gt;right thing&lt;/u&gt;&lt;/strong&gt;?&quot;&lt;/p&gt;

&lt;p class=&quot;callout-sep&quot;&gt;not&lt;/p&gt;

&lt;p class=&quot;callout&quot;&gt;&quot;did it build the &lt;strong&gt;&lt;u&gt;thing right&lt;/u&gt;&lt;/strong&gt;?&quot;&lt;/p&gt;

&lt;h2&gt;Part 4: The turn (top down)&lt;/h2&gt;
&lt;p&gt;Everything up to this point has been bottom up. Start from the smallest unit (a task) and keep making it bigger. Task → project → project of projects. And after SSP, it kind of stopped mattering how big the project is. As long as you know what it should look like at the end, with a proper spec and a mock, it gets built. It might run out of usage credits eventually, but that can be solved with a few extra accounts if need be.&lt;/p&gt;
&lt;p&gt;Which moved the bottleneck. Again. Now it&amp;#39;s the spec itself.&lt;/p&gt;
&lt;p&gt;I know the problems I have. What&amp;#39;s slow is me turning a problem into a strategy, and a strategy into the right thing to build. My ability to do that is now so much slower than the ability to execute that it&amp;#39;s almost funny.&lt;/p&gt;
&lt;p&gt;So now I&amp;#39;m going the other way. Instead of taking a spec and breaking it down into tasks, I take a &lt;strong&gt;goal&lt;/strong&gt; and break it down into the spec. And I mean a real-life goal (grow the business, land customers), not a development one.&lt;/p&gt;
&lt;figure class=&quot;diagram&quot;&gt;&lt;svg viewBox=&quot;0 0 760 272&quot; role=&quot;img&quot; aria-label=&quot;Goalset works down to the spec; the Coding Stack takes it from there&quot; font-size=&quot;13&quot;&gt;&lt;defs&gt;&lt;marker id=&quot;dir-arrow&quot; viewBox=&quot;0 0 10 10&quot; refX=&quot;9&quot; refY=&quot;5&quot; markerWidth=&quot;6&quot; markerHeight=&quot;6&quot; orient=&quot;auto-start-reverse&quot;&gt;&lt;path d=&quot;M0 0L10 5L0 10z&quot; fill=&quot;var(--cds-chart-axis)&quot;&gt;&lt;/path&gt;&lt;/marker&gt;&lt;/defs&gt;&lt;text x=&quot;24&quot; y=&quot;30&quot; font-size=&quot;15&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset works down to the spec; the Coding Stack takes it from there&lt;/text&gt;&lt;g&gt;&lt;path d=&quot;M24 104V96H488V104&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;text x=&quot;256&quot; y=&quot;84&quot; text-anchor=&quot;middle&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset (new): goal → spec&lt;/text&gt;&lt;g&gt;&lt;rect x=&quot;24&quot; y=&quot;120&quot; width=&quot;104&quot; height=&quot;42&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;2.5&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;76&quot; y=&quot;139&quot; text-anchor=&quot;middle&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goal&lt;/text&gt;&lt;text x=&quot;76&quot; y=&quot;154&quot; text-anchor=&quot;middle&quot; font-size=&quot;10.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;real-world&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;144&quot; y=&quot;120&quot; width=&quot;104&quot; height=&quot;42&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;196&quot; y=&quot;146&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Subgoals&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;264&quot; y=&quot;120&quot; width=&quot;104&quot; height=&quot;42&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;316&quot; y=&quot;146&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Strategies&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;384&quot; y=&quot;120&quot; width=&quot;104&quot; height=&quot;42&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;2.5&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;436&quot; y=&quot;146&quot; text-anchor=&quot;middle&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Spec + mock&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;504&quot; y=&quot;120&quot; width=&quot;104&quot; height=&quot;42&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;556&quot; y=&quot;146&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Projects&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;624&quot; y=&quot;120&quot; width=&quot;104&quot; height=&quot;42&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;676&quot; y=&quot;146&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Tasks → code&lt;/text&gt;&lt;/g&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;path d=&quot;M128 141H142&quot; marker-end=&quot;url(#dir-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M248 141H262&quot; marker-end=&quot;url(#dir-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M368 141H382&quot; marker-end=&quot;url(#dir-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M488 141H502&quot; marker-end=&quot;url(#dir-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M608 141H622&quot; marker-end=&quot;url(#dir-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M76 186V166&quot; marker-end=&quot;url(#dir-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M436 186V166&quot; marker-end=&quot;url(#dir-arrow)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;text x=&quot;76&quot; y=&quot;202&quot; text-anchor=&quot;middle&quot; font-size=&quot;12&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Now I start here&lt;/text&gt;&lt;text x=&quot;76&quot; y=&quot;218&quot; text-anchor=&quot;middle&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;not a dev goal&lt;/text&gt;&lt;text x=&quot;436&quot; y=&quot;202&quot; text-anchor=&quot;middle&quot; font-size=&quot;12&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;I used to start my dev work here&lt;/text&gt;&lt;g&gt;&lt;path d=&quot;M384 222V230H728V222&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;text x=&quot;556&quot; y=&quot;252&quot; text-anchor=&quot;middle&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Coding Stack: spec → code&lt;/text&gt;&lt;/svg&gt;&lt;figcaption&gt;The other way · goal → spec, then spec → code&lt;/figcaption&gt;&lt;/figure&gt;

&lt;h3&gt;Goalset: the brain of the operation&lt;/h3&gt;
&lt;p&gt;I&amp;#39;m aware I&amp;#39;m not the first person to try this. Back in 2023, BabyAGI and AutoGPT tried turning an objective into a task list and mostly got stuck in loops. Today there are idea validators that rank your riskiest assumptions, enterprise strategy tools, and well-funded startups (one raised $30M) trying to run your whole company with AI agents. But they each do half the job. The validators tell you what to test and stop there. The &amp;quot;AI runs your company&amp;quot; tools execute fast but don&amp;#39;t validate. They&amp;#39;ll build and market whatever they&amp;#39;re given, whether or not anyone wants it. I haven&amp;#39;t seen anything that does the whole loop: goal → strategy → test the assumptions → build → learn → re-plan. Which makes sense. The people who can build something this complex are devs who have been using these tools every day for the last 2 years. And most of them aren&amp;#39;t concerned this deeply with the ideation phase yet, because they&amp;#39;re still busy building their dev workflows. I talked to a dev friend the other day who had never even heard of Superpowers. Some devs are still using Copilot.&lt;/p&gt;
&lt;p&gt;The idea is that you give Goalset one big goal. It keeps a living plan: what has to be true, which strategies could get there, what evidence is missing, which small projects to run next, and what only you can do. Every goal gets stress-tested: which assumptions are riskiest, which ones are cheap to test right now, and which strategy the evidence actually supports. This is basically what I am doing every day as a founder.&lt;/p&gt;
&lt;p&gt;And then it keeps going after the plan is made, which is what makes it a brain and not a to-do list. Say the strategy for sub-goal A1 didn&amp;#39;t account for something a customer told me. That information flows back up, and every strategy, task and piece of research that depended on it gets redone. It always knows what the options are, what the assumptions are, and what we&amp;#39;ve learned.&lt;/p&gt;
&lt;p&gt;People have been trying to write this kind of thinking down forever (Ray Dalio&amp;#39;s &lt;em&gt;Principles&lt;/em&gt; being the obvious one). The difference is that now the written-down process can actually be &lt;em&gt;run&lt;/em&gt;. As deep as it needs to go, by something that never gets tired of asking &amp;quot;ok but why would that be true?&amp;quot;&lt;/p&gt;
&lt;h3&gt;What it does vs. what I do&lt;/h3&gt;
&lt;p&gt;When Goalset decides something needs to be built, it doesn&amp;#39;t need me to write a spec. It hands the project to SSP, SSP hands it to Superpowers, and it gets built. Before I even know what needs to be built, it&amp;#39;s already built.&lt;/p&gt;
&lt;p&gt;But some things AI can&amp;#39;t do yet. Talking to customers. Getting information out of the one person who has it. Being in the right room with the right people. (Eventually it&amp;#39;ll make the calls and send the emails too. Not yet.) So Goalset splits everything into two lists: what the machine does, and what I do. My list shows up with everything already prepared. I go do it, tell Goalset what I learned, and the whole plan updates.&lt;/p&gt;
&lt;figure class=&quot;diagram&quot;&gt;&lt;svg viewBox=&quot;0 0 760 580&quot; role=&quot;img&quot; aria-label=&quot;Goalset breaks a goal down into projects, then each project goes to AI or to me&quot; font-size=&quot;13&quot;&gt;&lt;defs&gt;&lt;marker id=&quot;split-arrow&quot; viewBox=&quot;0 0 10 10&quot; refX=&quot;9&quot; refY=&quot;5&quot; markerWidth=&quot;6&quot; markerHeight=&quot;6&quot; orient=&quot;auto-start-reverse&quot;&gt;&lt;path d=&quot;M0 0L10 5L0 10z&quot; fill=&quot;var(--cds-chart-axis)&quot;&gt;&lt;/path&gt;&lt;/marker&gt;&lt;/defs&gt;&lt;text x=&quot;24&quot; y=&quot;30&quot; font-size=&quot;15&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset breaks a goal down into projects, then each project goes to AI or to me&lt;/text&gt;&lt;g&gt;&lt;rect x=&quot;48&quot; y=&quot;48&quot; width=&quot;688&quot; height=&quot;84&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;2&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;64&quot; y=&quot;70&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset&lt;/text&gt;&lt;rect x=&quot;64&quot; y=&quot;80&quot; width=&quot;120&quot; height=&quot;36&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;124&quot; y=&quot;102&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goal&lt;/text&gt;&lt;rect x=&quot;224&quot; y=&quot;80&quot; width=&quot;140&quot; height=&quot;36&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;294&quot; y=&quot;102&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Subgoals&lt;/text&gt;&lt;rect x=&quot;404&quot; y=&quot;80&quot; width=&quot;200&quot; height=&quot;36&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;504&quot; y=&quot;102&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Strategies + research&lt;/text&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;path d=&quot;M184 98H222&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M364 98H402&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;272&quot; y=&quot;156&quot; width=&quot;160&quot; height=&quot;36&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;2&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;352&quot; y=&quot;178&quot; text-anchor=&quot;middle&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Projects&lt;/text&gt;&lt;/g&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;path d=&quot;M504 116V144H352V154&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M312 192V206H194V222&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M392 192V206H550V222&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M194 512V540&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M550 512V540H28V90H46&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;48&quot; y=&quot;224&quot; width=&quot;292&quot; height=&quot;288&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;64&quot; y=&quot;250&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;AI action → SSP&lt;/text&gt;&lt;rect x=&quot;64&quot; y=&quot;264&quot; width=&quot;260&quot; height=&quot;40&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;194&quot; y=&quot;288&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;SSP: plans, orchestrates, tests&lt;/text&gt;&lt;rect x=&quot;64&quot; y=&quot;320&quot; width=&quot;260&quot; height=&quot;40&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;194&quot; y=&quot;344&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Superpowers: builds each project&lt;/text&gt;&lt;rect x=&quot;64&quot; y=&quot;376&quot; width=&quot;260&quot; height=&quot;40&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;194&quot; y=&quot;400&quot; text-anchor=&quot;middle&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Claude Code (and other tools)&lt;/text&gt;&lt;rect x=&quot;64&quot; y=&quot;444&quot; width=&quot;260&quot; height=&quot;40&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-reference-tint)&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;4 3&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;194&quot; y=&quot;468&quot; text-anchor=&quot;middle&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Built software&lt;/text&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;path d=&quot;M194 304V318&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M194 360V374&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;path d=&quot;M194 416V442&quot; marker-end=&quot;url(#split-arrow)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;/g&gt;&lt;g&gt;&lt;rect x=&quot;364&quot; y=&quot;224&quot; width=&quot;372&quot; height=&quot;288&quot; rx=&quot;8&quot; fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;380&quot; y=&quot;250&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Human action → task timeline (like Linear)&lt;/text&gt;&lt;g&gt;&lt;line x1=&quot;528&quot; x2=&quot;528&quot; y1=&quot;280&quot; y2=&quot;480&quot; stroke=&quot;var(--cds-chart-grid)&quot;&gt;&lt;/line&gt;&lt;line x1=&quot;576&quot; x2=&quot;576&quot; y1=&quot;280&quot; y2=&quot;480&quot; stroke=&quot;var(--cds-chart-grid)&quot;&gt;&lt;/line&gt;&lt;line x1=&quot;624&quot; x2=&quot;624&quot; y1=&quot;280&quot; y2=&quot;480&quot; stroke=&quot;var(--cds-chart-grid)&quot;&gt;&lt;/line&gt;&lt;line x1=&quot;672&quot; x2=&quot;672&quot; y1=&quot;280&quot; y2=&quot;480&quot; stroke=&quot;var(--cds-chart-grid)&quot;&gt;&lt;/line&gt;&lt;line x1=&quot;720&quot; x2=&quot;720&quot; y1=&quot;280&quot; y2=&quot;480&quot; stroke=&quot;var(--cds-chart-grid)&quot;&gt;&lt;/line&gt;&lt;line x1=&quot;528&quot; x2=&quot;720&quot; y1=&quot;280&quot; y2=&quot;280&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/line&gt;&lt;/g&gt;&lt;text x=&quot;528&quot; y=&quot;272&quot; text-anchor=&quot;middle&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;now&lt;/text&gt;&lt;text x=&quot;720&quot; y=&quot;272&quot; text-anchor=&quot;middle&quot; font-size=&quot;11&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;later&lt;/text&gt;&lt;text x=&quot;380&quot; y=&quot;310&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Talk to customers&lt;/text&gt;&lt;rect x=&quot;532&quot; y=&quot;298&quot; width=&quot;78&quot; height=&quot;16&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.22&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;380&quot; y=&quot;350&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Meet the right people&lt;/text&gt;&lt;rect x=&quot;566&quot; y=&quot;338&quot; width=&quot;94&quot; height=&quot;16&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.22&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;380&quot; y=&quot;390&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Get info from someone&lt;/text&gt;&lt;rect x=&quot;600&quot; y=&quot;378&quot; width=&quot;90&quot; height=&quot;16&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.22&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;380&quot; y=&quot;430&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Hire people&lt;/text&gt;&lt;rect x=&quot;620&quot; y=&quot;418&quot; width=&quot;96&quot; height=&quot;16&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.22&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;380&quot; y=&quot;470&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Report what I learned&lt;/text&gt;&lt;rect x=&quot;660&quot; y=&quot;458&quot; width=&quot;56&quot; height=&quot;16&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.22&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;/rect&gt;&lt;/g&gt;&lt;text x=&quot;392&quot; y=&quot;562&quot; text-anchor=&quot;middle&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;What got built and what I learned flow back up; Goalset re-plans&lt;/text&gt;&lt;/svg&gt;&lt;figcaption&gt;The split · AI action vs. human action&lt;/figcaption&gt;&lt;/figure&gt;

&lt;p&gt;And to be clear, it doesn&amp;#39;t run without me. It never picks a strategy, spends money or contacts anyone on its own. Those decisions sit in my inbox with a recommendation and a &amp;quot;why is this here?&amp;quot; that traces all the way back to the goal. My job now is basically being an advisor to the CEO. Does this strategy make sense? Is it missing something? Is it true?&lt;/p&gt;
&lt;p&gt;So I wake up and the strategy is already set. It&amp;#39;s been updated overnight with whatever came in: replies from some automated outreach, or the notes I put in last night from my meetings. The software updates are already built. And I go talk to the people the system tells me to talk to.&lt;/p&gt;
&lt;p&gt;The only human work left is judgment and relationships.&lt;/p&gt;
&lt;p&gt;For now.&lt;/p&gt;
&lt;p&gt;Every card on my board is a task AI can&amp;#39;t do &lt;em&gt;yet&lt;/em&gt;. And that list only goes in one direction. Today I&amp;#39;m the one sending the emails and making the calls. Soon that moves to the AI side too. Eventually it&amp;#39;ll handle the human actions as well. Even setting the goal moves off my plate, to the next thing after Goalset: an agent that does pure idea generation. The only thing left for me is reviewing the strategy. Same pattern as every layer below it: write down what the human keeps doing, then hand it to the machine.&lt;/p&gt;
&lt;figure class=&quot;diagram&quot;&gt;&lt;svg viewBox=&quot;0 0 760 630&quot; role=&quot;img&quot; aria-label=&quot;Over time, the human list shrinks&quot; font-size=&quot;12&quot;&gt;&lt;defs&gt;&lt;marker id=&quot;time-arrow&quot; viewBox=&quot;0 0 10 10&quot; refX=&quot;9&quot; refY=&quot;5&quot; markerWidth=&quot;6&quot; markerHeight=&quot;6&quot; orient=&quot;auto-start-reverse&quot;&gt;&lt;path d=&quot;M0 0L10 5L0 10z&quot; fill=&quot;var(--cds-chart-axis)&quot;&gt;&lt;/path&gt;&lt;/marker&gt;&lt;/defs&gt;&lt;text x=&quot;24&quot; y=&quot;30&quot; font-size=&quot;15&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Over time, the human list shrinks&lt;/text&gt;&lt;text x=&quot;24&quot; y=&quot;50&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;Dashed = not built yet&lt;/text&gt;&lt;g&gt;&lt;text x=&quot;24&quot; y=&quot;92&quot; font-size=&quot;14&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Today&lt;/text&gt;&lt;rect x=&quot;24&quot; y=&quot;108&quot; width=&quot;220&quot; height=&quot;82&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;40&quot; y=&quot;132&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;154&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Subgoals + strategies&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;172&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Research&lt;/text&gt;&lt;rect x=&quot;24&quot; y=&quot;198&quot; width=&quot;220&quot; height=&quot;90&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;40&quot; y=&quot;222&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Coding Stack&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;242&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;SSP + Superpowers + CC&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;264&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Builds all the software&lt;/text&gt;&lt;rect x=&quot;24&quot; y=&quot;296&quot; width=&quot;220&quot; height=&quot;262&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;40&quot; y=&quot;320&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Me&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;342&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Come up with ideas&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;360&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Set the goal&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;378&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review the mocks&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;396&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review the strategy&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;414&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Talk to customers&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;432&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Meet the right people&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;450&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Get info from someone&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;468&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Send the emails, make the calls&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;486&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Hire people&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;504&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Raise money&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;522&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Approve spend&lt;/text&gt;&lt;text x=&quot;40&quot; y=&quot;540&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Sign contracts&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;text x=&quot;270&quot; y=&quot;92&quot; font-size=&quot;14&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Soon&lt;/text&gt;&lt;rect x=&quot;270&quot; y=&quot;108&quot; width=&quot;220&quot; height=&quot;148&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;5 4&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;286&quot; y=&quot;132&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;154&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Subgoals + strategies&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;172&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Research&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;190&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Send the emails&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;208&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Make the calls&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;226&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Find candidates to hire&lt;/text&gt;&lt;rect x=&quot;270&quot; y=&quot;264&quot; width=&quot;220&quot; height=&quot;90&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;286&quot; y=&quot;288&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Coding Stack&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;308&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;SSP + Superpowers + CC&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;330&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Builds all the software&lt;/text&gt;&lt;rect x=&quot;270&quot; y=&quot;362&quot; width=&quot;220&quot; height=&quot;196&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;5 4&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;286&quot; y=&quot;386&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Me&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;408&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Come up with ideas&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;426&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Set the goal&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;444&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review the mocks&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;462&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review the strategy&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;480&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Meet the right people&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;498&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Hire people&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;516&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Raise money&lt;/text&gt;&lt;text x=&quot;286&quot; y=&quot;534&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Approve spend&lt;/text&gt;&lt;/g&gt;&lt;g&gt;&lt;text x=&quot;516&quot; y=&quot;92&quot; font-size=&quot;14&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Eventually&lt;/text&gt;&lt;rect x=&quot;516&quot; y=&quot;108&quot; width=&quot;220&quot; height=&quot;98&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-2)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-2)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;5 4&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;532&quot; y=&quot;132&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Idea agent&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;154&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Comes up with ideas&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;172&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Sets the goal&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;192&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;the next thing after Goalset&lt;/text&gt;&lt;rect x=&quot;516&quot; y=&quot;214&quot; width=&quot;220&quot; height=&quot;156&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-1)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-1)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;5 4&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;532&quot; y=&quot;238&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Goalset&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;260&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Subgoals + strategies&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;278&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;All of the human actions too&lt;/text&gt;&lt;rect x=&quot;516&quot; y=&quot;378&quot; width=&quot;220&quot; height=&quot;90&quot; rx=&quot;8&quot; fill=&quot;var(--cds-chart-categorical-3)&quot; fill-opacity=&quot;0.12&quot; stroke=&quot;var(--cds-chart-categorical-3)&quot; stroke-width=&quot;1.5&quot; stroke-dasharray=&quot;none&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;532&quot; y=&quot;402&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Coding Stack&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;422&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;SSP + Superpowers + CC&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;444&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Builds all the software&lt;/text&gt;&lt;rect x=&quot;516&quot; y=&quot;476&quot; width=&quot;220&quot; height=&quot;82&quot; rx=&quot;8&quot; fill=&quot;none&quot; fill-opacity=&quot;1&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot; stroke-dasharray=&quot;5 4&quot;&gt;&lt;/rect&gt;&lt;text x=&quot;532&quot; y=&quot;500&quot; font-size=&quot;13&quot; font-weight=&quot;600&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Me&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;522&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review the goal&lt;/text&gt;&lt;text x=&quot;532&quot; y=&quot;540&quot; fill=&quot;var(--cds-text-primary)&quot;&gt;Review the strategy&lt;/text&gt;&lt;/g&gt;&lt;g fill=&quot;none&quot; stroke=&quot;var(--cds-chart-axis)&quot; stroke-width=&quot;1.25&quot;&gt;&lt;path d=&quot;M24 590H734&quot; marker-end=&quot;url(#time-arrow)&quot;&gt;&lt;/path&gt;&lt;/g&gt;&lt;text x=&quot;734&quot; y=&quot;612&quot; text-anchor=&quot;end&quot; font-size=&quot;11.5&quot; fill=&quot;var(--cds-text-secondary)&quot;&gt;over time&lt;/text&gt;&lt;/svg&gt;&lt;figcaption&gt;The human list over time · today, soon, eventually&lt;/figcaption&gt;&lt;/figure&gt;

&lt;h2&gt;Part 5: Where this goes&lt;/h2&gt;
&lt;p&gt;If this works (big if, I know), three things happen.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;First, my startup.&lt;/strong&gt; I am building a company right now. Goalset should be able to tell us what to build next, where the next opportunity to grow is, and which assumptions we&amp;#39;re betting on without even realizing it. The worst place a startup can be is in a rut, not knowing what to do next. Or, even worse, working on the wrong thing entirely. A system that always knows the next move keeps you out of it. My startup is basically my first test, and my calibrator. Because here&amp;#39;s the catch: the higher up the stack you go, the longer it takes to validate that the system actually works, and that grows exponentially. Claude Code takes a moment to test. Superpowers takes an hour or two. SSP takes days. Goalset is going to take months.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Second, more companies.&lt;/strong&gt; I take the proceeds from my startup and point the same system at new goals, at a rate I just couldn&amp;#39;t before. Each company only needs a couple of people, and they don&amp;#39;t have to be visionaries. They just have to be good at the human list: talking to customers, building relationships, reporting back what they learned. Theoretically you could hire a contractor to be the &amp;quot;founder,&amp;quot; have them wake up every day, do the tasks, and feed what they learn back in. And when something we assumed turns out not to be true, the system re-plans: new strategy, new tasks, new research, everything downstream redone. At that point it&amp;#39;s less a company and more a VC firm with a brain.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Third, the world.&lt;/strong&gt; If it works for me, it works for anyone who builds the same scaffolding. And then it gets very crazy, very fast. When ideation is as cheap as execution, the number of companies stops being limited by people having good ideas. I could see a world where there are basically no more companies left to build.&lt;/p&gt;
&lt;p&gt;That was honestly my fear about OpenAI and Anthropic. That they&amp;#39;d build this and there&amp;#39;d be nothing left for the rest of us. I don&amp;#39;t think so anymore. They&amp;#39;re working on the models, and I don&amp;#39;t think a stronger model is going to close this gap on its own. At every rung of this ladder, what closed the gap wasn&amp;#39;t intelligence. It was structure: a written-down record of where the model goes wrong, what to ask before acting, and where the human is actually needed. A smarter model makes each rung cheaper. It doesn&amp;#39;t build the rungs.&lt;/p&gt;
&lt;p&gt;In &lt;a href=&quot;/ai-coding-guide-retirement-is-here/&quot;&gt;February&lt;/a&gt; I predicted we&amp;#39;d go from relying on AI for execution to relying on it for ideation. I said it would take a decade. Eight months later, I&amp;#39;m already starting to get close.&lt;/p&gt;
&lt;p&gt;We&amp;#39;ll see what happens.&lt;/p&gt;
&lt;p&gt;Ciao, I&amp;#39;ll see y&amp;#39;all on the moon (still).&lt;/p&gt;
&lt;p&gt;Maxwell&lt;/p&gt;
</description></item>
<item><title>Reality as a Currency</title><link>https://maxwellnewman.com/reality-as-a-currency/</link><guid>https://maxwellnewman.com/reality-as-a-currency/</guid><pubDate>Thu, 16 Apr 2026 00:00:00 GMT</pubDate><description>&lt;p&gt;This morning I watched an interview with Jensen Huang, the CEO of NVIDIA. The last question he was asked in the interview was supposed to be a fun closing one: &amp;quot;If AI didn&amp;#39;t happen, what would NVIDIA be working on?&amp;quot; I thought his answer was the most profound part of the interview.&lt;/p&gt;
&lt;p&gt;He explained that NVIDIA&amp;#39;s success is tied to Moore&amp;#39;s Law (the number of transistors on a chip doubles roughly every two years). NVIDIA is positioned to continually increase compute power. In other words, even if AI never emerged, NVIDIA would still be incredibly successful because the world was already trending toward greater computing power—whether in gaming, graphics, or other domains.&lt;/p&gt;
&lt;p&gt;In other words, the success of his company was not even up for debate, because he had piggybacked off of Moore&amp;#39;s Law, which was &amp;quot;truth&amp;quot;.&lt;/p&gt;
&lt;p&gt;That got me thinking about my own company and how we position ourselves. I was struggling to figure out how to get customers on board. How can I get them to buy my product when they won&amp;#39;t even listen to my pitch? And then I realized: whether it&amp;#39;s convincing a new employee to join your company, selling a new product to someone you&amp;#39;ve never met, or even getting an investor on board with NVIDIA, it all comes down to this: people must see your reality and think it is even more accurate than their own version of reality. Or at least adopt pieces of it. The more they adopt the more they trust you. This becomes a compounding effect.&lt;/p&gt;
&lt;p&gt;You have to show the future as you see it—make others believe the world is inevitably moving in that direction and that they are better off for knowing it. Once they trust your version of reality, they&amp;#39;ll trust where you are investing your efforts and invest themselves—whether that&amp;#39;s their time, money, or energy. This is why pitch decks work. You&amp;#39;re not pitching a product, you&amp;#39;re pitching a framework &amp;amp; prediction of the world.&lt;/p&gt;
&lt;p&gt;The cool thing is, it&amp;#39;s far easier to get someone to listen to your vision of how their future might unfold than to immediately persuade them to act. You can put that in a cold email. Once they see your reality and how it might shape theirs, they&amp;#39;re far more likely to do what you want.&lt;/p&gt;
&lt;p&gt;There&amp;#39;s also the dangerous side of this, since this phenomenon is often done unintentionally. Social media (reels) has entire sections of the population brainwashed into believing certain views of reality subconsciously — even on subjects they never cared about. Gender wars, income inequality, political views; your stance on them all stems from a particular POV that you were fed at some point that made more sense than whatever model (or lack of model) you had previously. So the real power of influence in this world is how good you are at convincing people of your reality and spreading it.&lt;/p&gt;
&lt;p&gt;I think writing is a great method of practice to create this currency intentionally. Writing gives you the opportunity to refine your point of view, distill it down into concrete ideas, and consider the objections ahead of time. It&amp;#39;s why I write these posts. It helps me see truth more clearly than I did before. The act of sharing forces me to refine it. Paul Graham also makes a prediction that there will be a split between those who write and those who do not. The &amp;quot;&lt;a href=&quot;https://paulgraham.com/writes.html&quot;&gt;Writes and Write-Nots&lt;/a&gt;&amp;quot;. I believe writing is the greatest protection against the brainwashing phenomenon and forced thinking.&lt;/p&gt;
&lt;p&gt;Check in. In the last few moments after reading this: are you distrusting social media and what you believe a little bit more? Are you noticing things that you believed automatically? Are you thinking about different things you could potentially write about? Did I convince you that writing is good? If so, you just bought a little bit more into my reality. From this position it&amp;#39;s probably a lot easier to offer you my writing course on how you can alter other people&amp;#39;s reality too for as little as $9.99/mo that teaches you how… blah blah blah.&lt;/p&gt;
&lt;p&gt;Do good,&lt;/p&gt;
&lt;p&gt;Maxwell&lt;/p&gt;
</description></item>
<item><title>AI Coding Guide: Retirement is here.</title><link>https://maxwellnewman.com/ai-coding-guide-retirement-is-here/</link><guid>https://maxwellnewman.com/ai-coding-guide-retirement-is-here/</guid><pubDate>Wed, 25 Feb 2026 00:00:00 GMT</pubDate><description>&lt;p&gt;I have been using Claude Code since day 0. (Actually before day 0, I set up an MCP to do what Claude Code did before they released a product for it). The type of code orchestration Anthropic has had was half a year ahead of its time. (See the &lt;a href=&quot;https://www.instagram.com/reel/DMY8IGYMNB2/&quot;&gt;video I made 32w ago&lt;/a&gt; on this topic). It was the only product on the market for a solid 6+ months that actually sped up engineering an additional 5-10x from the traditional AI tools. The rest of the tech world has recently caught on (both in product &amp;amp; investing valuations) with Codex being really the only other orchestrator that has been able to keep up. (Google apparently has Gemini Code Assist that I have not tried but is allegedly pretty good too).&lt;/p&gt;
&lt;p&gt;The result: over the last year, AI has condensed months of software engineering work into a few days.&lt;/p&gt;
&lt;p&gt;The next question I (and others) ask is:&lt;/p&gt;
&lt;p&gt;How do I condense those few days into a couple hours?&lt;/p&gt;
&lt;p&gt;.&lt;br&gt;.&lt;br&gt;.&lt;/p&gt;
&lt;p&gt;(Then minutes?)&lt;/p&gt;
&lt;p&gt;I feel like 99% of the world doesn&amp;#39;t truly understand how much this is affecting our world. Even some software engineers that are out of the loop are just now waking up to Claude Code / Codex over poorer options like Cursor or Windsurf. &lt;strong&gt;*cough*&lt;/strong&gt; META engineers &lt;strong&gt;*cough*&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;But it takes time for adoption to spread, because a couple pioneers need to create their optimal systems, and then have those systems adopted by others as the &amp;quot;standard&amp;quot;. Even when the standard is set, people love to debate, and it takes weeks or months for them to even attempt a new standard and get up to speed. This is assuming there&amp;#39;s even a product for it. As I mentioned, I was already iterating with my own ghetto version of &amp;quot;Claude Code&amp;quot; before it was even released.&lt;/p&gt;
&lt;p&gt;However, I think that exact type of iteration will be what sets engineers apart in the next 1-5y.&lt;/p&gt;
&lt;p&gt;Until we are all self-employed.&lt;/p&gt;
&lt;p&gt;I believe I am currently experiencing the next exponential jump. I am not joking when I say I have made over a year (or two? or three? it&amp;#39;s hard to quantify accurately anymore) of traditional software progress in just 2 months.&lt;/p&gt;
&lt;h2&gt;Thesis&lt;/h2&gt;
&lt;p&gt;With the rise of execution becoming exponential, we&amp;#39;re hitting a point where planning and human understanding are becoming a bottleneck. There are so many things I know I need to do for my project / product, but the rate at which execution is being done (really f***ing fast), planning and understanding can seldom keep up. By the time I have learned and planned something I&amp;#39;m basically halfway done with execution. Not only that, from most engineers I know, I am constantly hearing &amp;quot;AI came up with a solution much better than what I could have in a fraction of the time.&amp;quot; So if I ended up doing the bulk of the planning myself without running it against another brain (artificial or meat) it probably wouldn&amp;#39;t even be that good.&lt;/p&gt;
&lt;p&gt;Despite AI-driven development being exciting, cutting edge and fast, it makes a ton of mistakes. It still takes quite a long time for more ambitious projects due to the nature of those mistakes. But prompting &amp;amp; prompting-organization have gotten so good (Claude superpowers), that we are now hitting the tipping point where the developer has become a bottleneck instead of an enabler.&lt;/p&gt;
&lt;p&gt;Right now the efficient, standard, AI dev process looks something like this:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;I come up with an idea / have a project to solve&lt;/li&gt;
&lt;li&gt;I plan out in my head, and/or an ERD/PRD (Engineering/product requirement doc) what needs to happen&lt;/li&gt;
&lt;li&gt;I break it down into parts&lt;/li&gt;
&lt;li&gt;I ask AI to solve these parts one by one, or I simply explain each piece out in a long-winded task list (and pray that AI doesn&amp;#39;t screw it up)&lt;/li&gt;
&lt;li&gt;Review, and go down the list&lt;/li&gt;
&lt;li&gt;Repeat 3-5 every 10-30 minutes for 8h a day&lt;/li&gt;
&lt;li&gt;Debug debug debug&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;Often though, you know what the issues are when they come up. They are pretty obvious and repeatable throughout different projects when doing code review. There&amp;#39;s so much context AI needs to have for execution that it will inevitably mess up if it is missing or misunderstanding a piece of implementation, which, if not caught early, can cascade into a ton of deeper issues down the line. Often (if you&amp;#39;re a semi-decent engineer) you will know where these major decision issues will be, and offer a bit more care to address them.&lt;/p&gt;
&lt;p&gt;However, if you knew exactly what mistakes could happen, and what your response should be for the next 100 iterations (you knew the questions it was going to ask you ahead of time, and you know what would be broken by AI ahead of time) you wouldn&amp;#39;t need to sit at the table and review everything that&amp;#39;s been output for hours.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;The goal:&lt;/strong&gt; remove the developer from the process as much as possible by front-loading and back-loading the prompting with the tools accessible (which will only get better). The developer is still needed, but we want to be mindful of the developer&amp;#39;s time and only require the developer on a need-to-know basis and need-to-execute basis. This becomes easy by essentially only involving the developer at the beginning &amp;amp; end of the dev process.&lt;/p&gt;
&lt;p&gt;This turns my 8h day of 10-30m intervals into 2-3 hour sessions over the course of a week, with 80+% of the time being used on implementation by the AI.&lt;/p&gt;
&lt;p&gt;But you have to be comfortable living in the dark. This is the first time AI knows the code better than I do, instead of the other way around (me teaching AI what it needs to know to execute correctly).&lt;/p&gt;
&lt;p&gt;With that said, here we go.&lt;/p&gt;
&lt;h2&gt;What I believe is currently the fastest form of development&lt;/h2&gt;
&lt;p&gt;First, some short-term trade-offs (since this is not standard yet).&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Positives:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;At least 5-10x faster than regular AI-driven development&lt;/li&gt;
&lt;li&gt;Does not require full understanding of implementation up front (can use cutting-edge tech without first learning about it)&lt;/li&gt;
&lt;li&gt;Lowers intelligence barrier-to-entry&lt;/li&gt;
&lt;li&gt;Allows you to implement more-elegant solutions than what you could have come up with on your own&lt;/li&gt;
&lt;li&gt;Less friction from idea → valid/smart/correct implementation&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;Negatives:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Large gaps in implementation knowledge&lt;ul&gt;
&lt;li&gt;&lt;em&gt;Hedged by the auditing step&lt;/em&gt;&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;With larger projects, AI is much less likely to connect all the pieces&lt;ul&gt;
&lt;li&gt;&lt;em&gt;Hedged by the auditing step&lt;/em&gt;&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;Deprecated code is more dangerous and becomes more annoying to find and manage&lt;ul&gt;
&lt;li&gt;&lt;em&gt;Hedged by the auditing step&lt;/em&gt;&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;It is slightly uncomfortable to let AI take the responsibility of thinking &amp;amp; organization instead of you&lt;ul&gt;
&lt;li&gt;&lt;em&gt;Hedged by Religion and Prayer&lt;/em&gt;&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Note: long-term, when the world catches up, and there&amp;#39;s billions (trillions?) of dollars behind this type of workflow, the negatives will be solved through more intensive worktrees than what I as a single individual have created below. I am building a company, and unfortunately don&amp;#39;t have the time to solve all of the problems I&amp;#39;ve seen (see Notes section below), but they are all 100% &amp;quot;easily&amp;quot; solvable.&lt;/p&gt;
&lt;h2&gt;Workflow&lt;/h2&gt;
&lt;h3&gt;Step 1: Product Design &lt;span class=&quot;aside&quot;&gt;3 hours – 1 week&lt;/span&gt;&lt;/h3&gt;
&lt;ul&gt;
&lt;li&gt;Use your favorite LLM chat (I use ChatGPT) to solidify what you want from your product, including research, understanding, engineering implementations at high levels, edge cases, issues, etc. (Product can be huge. The goal is to understand what it does and how it works at a HIGH level. Whatever you want, no limits. Actually try to stress test it. Seriously.)&lt;/li&gt;
&lt;li&gt;Note: False or incorrect assumptions here can be costly but not existential. Everyone has their own product designing phase and there&amp;#39;s plenty of material on this already so I won&amp;#39;t go into detail here, but the end goal of this step is to really, really, really know what you want, because we are extremely close to true one-shotting large, complex projects. That is why this timeline is so generous.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3&gt;Step 2: Engineering Design &lt;span class=&quot;aside&quot;&gt;2 hours&lt;/span&gt;&lt;/h3&gt;
&lt;ul&gt;
&lt;li&gt;Ask chat to summarize everything into a PRD with engineering specifics (no data-models). Ensure that it has implementation details for each step of the pipeline on how to implement&lt;/li&gt;
&lt;li&gt;Paste the PRD into a new context window. This window is now your &amp;quot;Project Manager&amp;quot;, not you.&lt;/li&gt;
&lt;li&gt;Open Claude and run &lt;code&gt;/superpowers:brainstorm&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;Paste in your PRD&lt;/li&gt;
&lt;li&gt;Claude will ask you various questions about implementation&lt;/li&gt;
&lt;li&gt;Don&amp;#39;t answer them! (unless you want to)&lt;/li&gt;
&lt;li&gt;Paste the questions to your ChatGPT Project Manager and get the answers&lt;/li&gt;
&lt;li&gt;Paste the response back to Claude&lt;/li&gt;
&lt;li&gt;Repeat until finished.&lt;/li&gt;
&lt;li&gt;Claude will generate a design document. Review the document (optional). Paste this doc into your project manager and ask it to review. Paste the response back into Claude 1-2 more times until solid.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3&gt;Step 3: Implementation Design &lt;span class=&quot;aside&quot;&gt;10 minutes&lt;/span&gt;&lt;/h3&gt;
&lt;ul&gt;
&lt;li&gt;Clear context in Claude Code&lt;/li&gt;
&lt;li&gt;Ask Claude to write the implementation doc for the design doc it created (&lt;code&gt;/superpowers:write-plan&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;Note: If your design is ESPECIALLY big, you may ask it / it may ask you to implement MULTIPLE implementation docs (I have had one design doc create 4 separate implementation docs, each with 30+ tasks.)&lt;/li&gt;
&lt;/ul&gt;
&lt;h3&gt;Step 4: AI Coding &lt;span class=&quot;aside&quot;&gt;1–24 hours waiting&lt;/span&gt;&lt;/h3&gt;
&lt;ul&gt;
&lt;li&gt;When complete, close Claude, re-open with &lt;code&gt;--dangerously-skip-permissions&lt;/code&gt; flag&lt;/li&gt;
&lt;li&gt;Ask it to execute superpowers for the first implementation doc&lt;/li&gt;
&lt;li&gt;Sit back and relax until complete&lt;/li&gt;
&lt;li&gt;Repeat step 4 for as many implementation docs as you have&lt;/li&gt;
&lt;li&gt;Each doc will take anywhere from a few hours to a few days depending on how large your project is&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Note: if you somehow realize you completely screwed up your PRD from the beginning (like I did), this is a good time to double / triple check that all of the assumptions you made are correct. Really you should have done this at the PRD stage, but at most, at this point you&amp;#39;ve only lost a few hours and a day or two of credits. If you realize the implementation you chose is not correct, repeat steps 1-4 using the design doc you created as an initial reference point.&lt;/p&gt;
&lt;h3&gt;Step 5: Code Audit &lt;span class=&quot;aside&quot;&gt;1–3 days&lt;/span&gt;&lt;/h3&gt;
&lt;p&gt;With any worthwhile or ambitious project, you need to make sure it actually worked. Claude will mess up connections between implementations (naturally — the context is WAY too large) and therefore it is imperative that you have a MEDIUM level understanding of the code (need not understand line by line, but you need to know what each function / step should be doing during execution).&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Ask Claude to explain to you how everything works, referencing implementation docs&lt;/li&gt;
&lt;li&gt;Ask Claude to create a workflow tree with relevant endpoints&lt;/li&gt;
&lt;li&gt;Run the pipeline / code / execution end to end manually once&lt;/li&gt;
&lt;li&gt;During this phase you may find bugs, solve them as you normally would with Claude&lt;/li&gt;
&lt;li&gt;When the bugs are fixed and the tree is created, now get granular and ask Claude to review each step one by one (traditional AI + human debugging. ex: &amp;quot;review step one in this pipeline&amp;quot;)&lt;/li&gt;
&lt;li&gt;By the end of this process you should have a much more solid understanding on what the hell AI actually did&lt;/li&gt;
&lt;li&gt;Create end-to-end tests for ensuring that the OUTCOME of the code gives you exactly what you want. When you understand the intricacies of the code as you walk through, have Claude create intermediate tests as well&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;At the end of this, you should have a working system in 1/100th of the time it would take with traditional engineering. At least a 10x gain from AI-enabled engineering. The reason for this is that debugging a correctly structured high-level implementation (this 5 step workflow) is much faster than building each piece of the structure to completion (current dev flow).&lt;/p&gt;
&lt;p&gt;Think of it like building a house. It&amp;#39;s a lot easier to build when all of the high-level skeleton is made. It&amp;#39;s not necessarily complete — but you know where the walls are, which room is where. You just need to connect the pieces on the inside.&lt;/p&gt;
&lt;p&gt;&lt;img src=&quot;/images/house-framing.jpg&quot; alt=&quot;A house with its wooden frame fully up but no walls yet&quot;&gt;&lt;/p&gt;
&lt;p&gt;Give it a try. I was able to code something I didn&amp;#39;t even fully understand until a month later. For the first time my understanding of the code was not a bottleneck to making progress. This has never been possible before in the history of engineering. Progress used to move as fast as the smartest person&amp;#39;s understanding moved. Now it can take a few leaps ahead and allow you to build code + understanding at the same time. You just have to be comfortable letting go of the wheel a little bit more.&lt;/p&gt;
&lt;p&gt;The reason this all works is because when you need to fix something, you can. It&amp;#39;s similar to working on a codebase in a company. You don&amp;#39;t need to understand the entire code, just the piece you are currently working on / fixing. And in many cases you don&amp;#39;t really need to know how it works, just what the inputs and outputs are and where they feed to/from.&lt;/p&gt;
&lt;p&gt;And trust me, you better get used to it, because it&amp;#39;s only going to get more automated from here. I put the next steps to make it even more automatic (see NOTE below). You can even do it yourself if you&amp;#39;re motivated enough to be a pioneer. Maybe you can be the next superpowers open source tool.&lt;/p&gt;
&lt;p&gt;This is already long enough. But it&amp;#39;s actually insane to me how quickly this is all changing. And just to make things even more enjoyable, as of 5 f***ing hours ago you can now run this entire workflow at the gym, at a party, on a walk. All of this can be done without even being at the computer.&lt;/p&gt;
&lt;p&gt;&lt;img src=&quot;/images/remote-control.png&quot; alt=&quot;Noah Zweben&amp;#39;s tweet announcing Claude Code Remote Control: start local sessions from the terminal, then continue them from your phone&quot;&gt;&lt;/p&gt;
&lt;p&gt;Insane. The above screenshot deserves its own post. This is fresh off the press and I can&amp;#39;t wait to work from the gym.&lt;/p&gt;
&lt;p&gt;Ciao, I&amp;#39;ll see y&amp;#39;all on the moon.&lt;/p&gt;
&lt;p&gt;Maxwell&lt;/p&gt;
&lt;hr&gt;
&lt;h2&gt;Notes&lt;/h2&gt;
&lt;ol&gt;
&lt;li&gt;Automating the PM → Claude connection is an easy next step. It&amp;#39;s what truly makes this an idea → code ready workflow. I actually think this can be done today.&lt;/li&gt;
&lt;li&gt;I believe there is a world where the code-auditing step can also be automated away soon. The only real requirement is to create workflows ahead of time to test in production. It&amp;#39;s a big build. It will require some intentional implementation there. This will be by far the biggest time unlock.&lt;/li&gt;
&lt;li&gt;Following that there would need to be some sort of orchestration for true black-box, end-to-end development testing to ensure what has been built truly works as intended. This is largely the biggest mountain, and is exponentially harder for a front-end implementation.&lt;/li&gt;
&lt;li&gt;Following that, we can then automate the PRD step. Build structure to help AI find the most optimal way to solve a problem and simply prompt the user on specifics on what they mean and to extract the context from the mind. This is the turning point where we focus less on optimizing workflow and more on extracting information from the developer.&lt;/li&gt;
&lt;li&gt;Following that, we start to go truly exponential. We automate ideas. This is where projects do not even need to be thought of before they are fully implemented. This is when the big AI giants begin building spin-off companies. This is what I am betting myself and my career on in the next decade.&lt;/li&gt;
&lt;/ol&gt;
</description></item>
</channel></rss>
