AI told me to quit the thing I love most

A live self-coaching session on using AI to find focus

AI + Web3 intelligence for builders and creators

Issue #009  ·  June 29, 2026  ·  theshipguild.com

This week: The most valuable thing AI did for us this week wasn't building faster — it was telling us what to stop doing. A live self-coaching session on using AI to find focus. Plus: the AI bubble Q keeps warning about started leaking, and Hyperliquid ripped while Bitcoin bled.

Section 1 · Livestream Recap

From Friday's "How We Use AI to Find Purpose", a live, unscripted self-coaching session

This week's stream was different. No news segment, no product demo, just Trav getting coached in real time on the hardest problem most builders never actually sit down to solve: what should I focus on? With agents now doing the work faster, the bottleneck isn't capacity anymore. It's direction. And the sharpest tool for finding it turned out to be AI itself, pointed inward.

Key takeaways from the stream:

Agents removed the capacity limit — so focus is the new bottleneck. Trav is now shipping a full portfolio website, a job-scraping agent, and an X account audit, all with AI. But Q's point cut through: if you use that new speed to pour energy into the wrong direction, you've just gotten more efficient at wasting time. Alignment has to come before output.

"Audit my account" is not enough — you'll get confidently false answers. Trav learned this the hard way. Hermes (not logged in) only saw his 2023 posts. Grok over-indexed on replies and inflated engagement. He had to use Claude to write a surgical prompt that cut through the noise and pulled the two months of content that actually mattered. The lesson: the quality of an AI audit is entirely determined by how you scope the data you feed it.

The AI told Trav to drop video editing. Q said the AI was wrong — and both were right. Based purely on metrics, the agent concluded Trav's video-editing content wasn't landing and should be cut. Q pushed back hard: the problem isn't the topic, it's the framing. Trav's video posts were about his personal journey, not useful to someone who wants to make videos with AI. Same skill, different packaging. Don't cut your unique edge, reframe it toward your audience's goal.

Two growth philosophies, one experiment to settle it. Trav's approach: grow a personal brand with broad "infotainment," let people funnel to Ship Guild naturally. Q's approach: pick the thing you're top-tier at, lean in all the way, show the receipts, and let the persona be the funnel. They didn't resolve it with opinions, they set up a two-week experiment and will report the data live.

Show the receipts — or don't claim the skill. Q's blunt framing on positioning as a social-media growth expert: "You're telling me you can grow my socials, but you have a hundred likes?" If you're going to sell growth, the proof has to be visible, handpicked spaces, stream results, real before/after numbers from the accounts you've actually moved.

"Don't ask the AI to be nice. Ask it for all the bad things, all the weak points. The phrase that worked for me was: 'please open my eyes in regards to [X].' It was the best feedback I've ever gotten from an AI."— Q, on running an honest external audit

The Big Idea: The internet is full of people telling you AI will help you do more. Almost nobody tells you the highest-leverage use is having it tell you what to stop. People are yes-men — they want to be nice. A well-prompted agent, explicitly told to be harsh, will ask you the questions you've been avoiding: What are you actually top-tier at? What can you do for years with no external reward? Where is your effort misaligned with your goal? Most creators never sit down to answer those, because it's uncomfortable and there's no dopamine in it. That discomfort is exactly where the direction lives. Use the agent to force the conversation with yourself you keep putting off.

What to do this week: Spin up a fresh agent with one job — audit your brand. Feed it your real, scoped data (not "look me up," but the specific posts, numbers, and pages that matter). Then run the two prompts below. Let it be brutal.

🪄 The External Audit Prompts: Please open my eyes in regards to [your project / account / brand]. Don't be nice, tell me every weak point, everything that's broken, everything misaligned with the persona I'm trying to build, and the hard questions I'm avoiding.

Then: Ask me the questions I need to answer to find my real focus and direction.

📺 Full stream: HERE  ·  🎵 TikTok clips: HERE

Section 2 · Method Spotlight

The External Audit — Point an Agent at Yourself, Then Ask It to Be Brutal

The single most useful AI workflow from this week's stream, and it costs nothing to try

The External Audit Workflow

Spin up an independent agent whose only job is to audit your brand, find what's broken, and ask you the hard questions

Most people use AI to produce. The move almost nobody makes is to use a separate, dedicated agent to evaluate, with no stake in making you feel good. It's the difference between an internal audit (you scrolling your own analytics, confirming what you already believe) and an external one (an outside perspective that surfaces the misalignment you can't see because you're inside it).

Internal Audit

External Audit

You review your own analytics and data

A fresh agent reviews your accounts as an outsider

Confirms what you already suspect

Surfaces blind spots and misalignment

Optimizes what you're already doing

Questions whether you should be doing it at all

How to run it:

1. Scope the data properly. Don't say "look me up." Agents without login access see stale or partial data on X (Trav's saw only 2023 posts). Feed it the specific, recent, relevant content, or use a prompt that forces it to cut past replies and inflated metrics to what actually matters. Trav used Grok with a ruthless prompt to grab and evaluate his last 2 months worth of content, view the entire process here👇

2. Explicitly tell it to be harsh. "Please open my eyes." "Don't be nice." "Give me every weak point." The default politeness setting is what makes most AI feedback useless.

3. Ask it to interrogate you. "Ask me the questions I need to answer to find my focus." Finding direction is a self-awareness exercise, if you don't have the right questions, have the agent generate them.

4. Bring human context to the verdict. The agent told Trav to drop video editing based on metrics alone. Q's years of context corrected it: reframe, don't cut. The audit is an input, not a ruling. You still decide.

⚠️ Honest caveat: An audit is only as good as the data you scope and the honesty you demand. Metrics-only agents miss context a human who knows you would catch — so treat the output as a sharp second opinion, not gospel. And one week of social-media data is too short to conclude much; A/B test with fixed conditions if you want real signal.

⚓ Accountability Challenge

The 2-Week Focus Experiment

Trav committed live to a two-week challenge: refocus his personal account entirely around growing on social media with AI, reframe every piece of content toward that single niche, polish his bio, and post consistently. Q audits the results live on next Friday's stream — with the data, not vibes.

Want to run it with him? Pick your one focus, reframe your next two weeks of content around it, and DM @shipguild your before/after. We'll feature the best transformations.

🔥 ENJOYED THIS ISSUE?

If Growth Bytes is part of your weekly stack, the best thing you can do is share it with one builder or creator who needs it.

🐦 X: HERE (formerly @web3matters_xyz)
📺 YouTube: HERE
🎵 TikTok: HERE
🌐 theshipguild.com

We drop every Monday. See you next week.
— Trav & Q, Ship Guild

Reply

or to participate.