I Automated the Word "No"

I've spent 36 months drowning in yeses.

Partnership calls that went nowhere. "Quick syncs" that ate 90 minutes and produced zero. Content angles I knew were mediocre but wrote anyway because I'd already said yes.

Saying no is expensive.

It costs attention, context, and emotional labor. So I defaulted to yes. My calendar became a graveyard of commitments I never actually chose.

Here's what most founders miss: Everyone's using AI to produce more. Blog posts. Social content. Email sequences. Faster, faster, faster.

They're accelerating the wrong work.

The real constraint isn't time, it's mental energy.

Every founder has maybe 4-5 hours of genuine 10/10 thinking per day. The question isn't "how do I do more?" It's "how do I protect those hours for work that actually matters?"

Two weeks ago, I built a system in Claude Code with access to my meeting history, content performance, and goals. It has one job: tell me what to ignore.

I call it The Judgment Filter. Here's exactly how I built it.

Did you recently see an ad for beehiiv in a newsletter? We’re running a short brand lift survey to understand what’s actually breaking through (and what’s not).

It takes about 20 seconds, the questions are super easy, and your feedback directly helps us improve how we show up in the newsletters you read and love.

If you’ve got a few moments, we’d really appreciate your insight.

Take the survey.

I was staring at a LinkedIn DM from a founder I'd never spoken to. Subject: "Quick question about your AI workflow."

The message hit all the beats: compliment, vague connection, ask for time. What it didn't have? An actual question. Just "15 minutes to pick your brain", which, in my experience, is code for:

This wasn’t from a subscriber with a specific question. Just a request for my time dressed up as curiosity.

Except it's never 15 minutes. It's the context switch before. The calendar Tetris. The 15 that becomes 35. The follow-up email. The "let's stay in touch" that becomes another thread to manage.

Instead of responding, I opened Claude Code and typed:

"Check this DM against my meeting history. What's my success rate with 'pick your brain' requests from people I don't know? Pattern match."

Three seconds later:

I stared at that for a minute.

Claude didn't just save me from the call. It saved me from the decision. From the guilt of saying no. From spending 10/10 energy on a 3/10 opportunity.

That's when I realized what I'd actually built.

Not a productivity tool. A taste externalization system.

Meaning: I stop spending founder brain on obvious no's. I only wake up for the weird yes's.

So I pushed it further.

A week later, The Judgment Filter caught something I would have missed.

A local business group invited me to be a featured speaker at their monthly meetup. "20-30 business owners, very engaged, great networking opportunity." They'd even offered a small speaking fee.

Old me would've said yes. I've said yes to these before. Show up, give a talk, shake hands, collect business cards, drive home.

You know how these actually go: half the room is insurance agents and "marketing consultants" who want to sell you something. You spend more time dodging pitches than making connections. The "great networking opportunity" is everyone trying to sell everyone else.

But I ran it through the filter first.

Here's the part that hit me: I would have said yes.

My gut said yes. It felt like "putting myself out there." The speaking fee felt like validation.

But the data said no. Claude saw what I couldn't, because Claude doesn't have an ego. Claude doesn't get dopamine from being asked to speak.

That was the moment I understood what this system actually does:

It's not about saving time. It's about catching the decisions I would have made wrong.

My taste failed. The filter caught it.

Once I trusted the system, I started running everything through it.

What I saved: roughly 15 hours of high-value energy that would have gone to 3/10 work.

Here's what I actually did with those 15 hours:

I wrote a newsletter that hit a 42% open rate. I had a convo for an online speaking opp for 2,400 growth marketers. I took a Wednesday afternoon to refine a feature for a product we’re launching.

The filter didn't just save time. It gave me back choice.

When Steve Jobs came back to Apple in 1997 he cut 70% of their product line.

The iPhone wasn't built by saying yes. It was built by a thousand strategic nos.

But here's what nobody tells you about the founder "no": It's not the decision that costs you. It's the consideration.

Every opportunity that reaches your desk (even the obvious nos) burns judgment. You have to load context. Weigh options. Navigate the social cost of declining.

The Judgment Filter doesn't just say no. It says no before the decision forms. Before you spend 10/10 energy on a 3/10 choice.

Now let me show you how to build it.

This was a series of "holy shit, this works" moments stacking on each other.

(Note: I use Claude Code, which runs in my computer's terminal. It looks like The Matrix, but it's actually cleaner and faster than clicking through a web browser. If you can copy-paste, you can use it.)

It started with my newsletter archive.

I use Beehiiv. Excellent for publishing, meh for analysis. I wanted Claude to know what content actually performed (not what I thought performed).

So I built an n8n workflow. Every time I publish a newsletter, it automatically saves the full post to a GitHub repo with the metrics embedded in the frontmatter. Open rates. Click rates. Subscriber count at send.

Now Claude can pull from that repo directly. When I ask "should I write about X?", it's not guessing. It's pattern-matching against 50+ real data points.

Then I gave Claude my social history.

I used ScrapeCreators to scrape every post I've ever published across X, Threads, everything. Dumped it all into my Obsidian vault.

Now Claude knows what hooks worked. What bombed. What got engagement vs. crickets. When I'm evaluating a new content angle, it can tell me: "This sounds like your May 15th post that got 12 likes. Your February post with a similar topic got 400. Here's the difference."

That's my own data telling me what to do.

But scraping my own data wasn't enough. Claude needed context on decisions, not just content.

So I connected my inbox and calendar.

I used the “Jean-Claude” custom MCP plugin (Model Context Protocol, it's how you give Claude access to external tools) that hooks directly into Gmail and Google Drive.

I just say "check this pitch in my inbox" and Claude pulls it.

Combined with my partnership log (a simple markdown file tracking every past partnership and outcome), Claude can pattern-match in real time.

Then I added my meeting transcripts.

I use Read.ai to transcribe every call. Those transcripts auto-save to Google Drive. I set up a simple workflow to pull them into my vault.

(Note: I’ve been saving my meeting transcripts to a vector database for the few years, but you can just pull your notes in and get 90% of the way there)

Now Claude has context on every meeting I've taken. When someone asks for a call, Claude can check: "You've had 3 calls with similar 'explore synergies' framing. Average length: 52 minutes. Outcomes: 0. Suggest declining."

Finally, I built Skills for repeatable workflows.

Claude Code has this feature called Skills, basically saved prompts that run complex workflows consistently. I built one for content evaluation. One for meeting screening. One for partnership review.

Now I don't even type the prompt. I just run /evaluate-content and Claude knows exactly what to check.

Structure is scaffolding. Here's my folder setup—copy it:

What feeds into each folder:

The CLAUDE.md files are where you encode taste.

You can put a file called CLAUDE.md in any folder, and Claude Code reads it automatically when working in that context. It's like a briefing before Claude starts.

Root CLAUDE.md (the master):

performance-data/CLAUDE.md:

decisions/CLAUDE.md:

These files look simple. But they took me 6 iterations to get right.

Here's what I learned: The key isn't telling Claude what to do. It's telling Claude how to think. The difference between "evaluate this partnership" and "find reasons to say no, then see if anything survives" is everything.

These prompts work because they're not asking Claude to decide. They're asking Claude to pattern-match against my own history.

There's a difference.

For content decisions:

For meeting requests:

For partnership pitches:

For opportunity evaluation:

Simple. Direct. Claude does the lookup. I get a verdict and a draft response.

Total setup time: About 6 hours.

ROI: Absurd. I'm making better calls with less brain.

You don't need my whole stack. Pick your entry point:

Open ChatGPT or Claude in your browser and paste this exact prompt:

Then paste the request. That's it. You just ran your first Judgment Filter.

Do this 5 times before your next "yes." You'll start seeing patterns.

Start with Path A. Graduate to Path B when you want persistent context that compounds over time.

Either way, you'll feel it immediately. That relief of not being the bottleneck.

The Judgment Filter isn't about doing less. It's about protecting your 10/10 energy for the work that actually matters.

Because the future isn't about doing more. It's about filtering better.

And the filter is already running for the founders who figured this out first.

As I was finishing this newsletter, Anthropic launched Cowork, an agentic tool that gives Claude access to a folder on your computer.

I haven't tested it yet, but the direction is right: removing the terminal barrier so non-devs can build systems like this with native connectors and a visual interface.

If you're not a Claude Code person, this might be your entry point. Same folder structure concepts, same CLAUDE.md approach, just without the command line.

Worth watching. I'll report back once I've put it through its paces.

🎁 Get rewarded for sharing! I’ve scraped and distilled every lesson from the entire AppMafia course (and built a custom AI trained on it all). This means you get the exact frameworks, insights, and viral strategies behind their $36M empire.

Invite your friends to join, and when they sign up, you’ll unlock my App Mafia GPT.

{{rp_personalized_text}}

Copy and paste this link: {{rp_refer_url}}