This AI agent built my creator strategy in 10 minutes

Your creator brief is lying to you.

You're about to drop $40K on sourcing, briefing, and production. Your team spent maybe an hour on the actual idea. That's not a strategy. That's a lottery ticket in a Google Doc.

Look at the doc in your inbox right now. "Authentic UGC. Three hook variants. Five lifestyle creators we like." Nobody pulled the real post-level outliers in your category. Nobody mapped your top buyer objection to a specific hook structure. Nobody rejected the generic plays before briefing.

It doesn't matter if you pay a creator with 10M followers. If the idea is flawed and the hook is generic, people swipe in three seconds. You just paid for production on a concept nobody validated.

So I built a kit that runs the strategy layer before sourcing. Pulls the real outliers in your niche. Maps objections to specific hook structures. Rejects the generic plays in writing, so the test plan has teeth before you spend a dollar.

Open source. Free. Here's how it works.

(scroll to the bottom for the link)

Every brand and every agency runs the same broken loop.

Step one, you decide you want creator content (or you hire an agency to run it for you). Step two, the "concepts deck" lands in your inbox. Step three, the deck is 80% creator comps someone already liked and 20% generic UGC advice pulled from a swipe file. Step four, the contract gets signed. Step five, sourcing starts. Step six, the brief says "authentic, hooks first, three variants."

If you're already running paid UGC with a system like this AI UGC army or AI posts like a 100k creator, you know the feeling. Execution is no longer the bottleneck. The bottleneck is the idea.

This isn't a creator problem. This isn't even a briefing problem.

The strategy layer never existed.

Nobody pulled the post-level outliers in the category. Nobody mapped the top buyer objection to a specific hook structure. Nobody decided which generic plays to reject. Nobody ran any of that before sourcing got spun up.

So the sourcing team optimized for the wrong thing. The briefing was generic because the concepts were generic. The creators filmed what looked like every other UGC spot on the platform. And the platform does not reward what looks like every other UGC spot on the platform.

Same philosophy behind the First 1000 Customers kit: decide the strategy layer first, let automation run the execution layer.

Creator Breakout Kit is an AI strategist that finds the creator concepts, hooks, and angles most likely to break out in a category (before you pay to source or produce a single video).

It runs a 6-stage pipeline. Each stage reads the last stage's artifacts. No stage hallucinates inputs. No stage ships without an honest evidence label.

The output is a Creator Breakout Brief. An 8-section document that reads like something a senior creator strategist with a week of research would hand you, except it runs in about 10 minutes.

It ships in two formats from one data source: a markdown file you can edit and feed into other agents, and a dark-terminal-aesthetic HTML file you can open in a browser and actually show a prospect without being embarrassed. Same score logic, same content, one parametric renderer. The formats can't drift from each other.

Works with Claude Code, OpenClaw, or any agent framework that can read skills.

Every run produces a brief with this exact structure.

Plus an appendix with brand context, hero products, audience, objections, and brand-safety guardrails.

Section 7 is the one that changes how the work feels. Most creator strategy docs are allergic to saying "no." This one explicitly rejects the 5 generic plays before it recommends anything. That single section is why clients read the brief and say "okay, you actually thought about this."

Every single concept in the slate carries four labels. This is what makes the brief feel ownable instead of vibes.

Then a score strip: Strategic fit · Breakout evidence · Production ease · Confidence. The confidence score is computed from the research mode and whether a real source post is attached. No research → Hypothesis only. Firecrawl + Virlo → Directional. Video intelligence attached → Strong directional.

The agent will not fake certainty. If the evidence isn't there, the label says so. That's the whole game.

Here's the gap, side by side.

Generic creator brief that ships today:

Creator Breakout Kit direction:

The first one buys you production. The second one buys you a test plan.

You could hand the second one to three different sourcing teams and get roughly the same creators back. You could hand the first one to three different sourcing teams and get three completely different decks. That's the signal that the strategy layer is missing.

The industry standard is find creators first. Pick based on audience fit or follower count. Brief a generic hook structure. Hope.

This kit flips it.

The creators don't change that much. What changes is what you ask them to do. And what you ask them to do is the entire difference between "this campaign worked" and "this campaign was vibes."

Every skill is a plain markdown file. Any agent framework that reads skills can run this.

This is the part that matters most. And it's the part every other "creator AI" tool gets wrong.

Most tools pull "top creators in the category." That's a follower-count lookup. It tells you who has the biggest audience, not who is making the content that's actually breaking out right now. (This is the same trap I flagged in the viral video kit: the platform doesn't reward follower count. It rewards pattern fit.)

Creator Breakout Kit uses post-level outlier logic. It goes into Virlo, pulls the posts that are outperforming a given creator's own baseline by a factor of 3x, 5x, 10x, and then looks at why. What's the hook shape. What's the mechanic. What's the buyer tension it's resolving. What's the filming style doing.

Outlier logic means the patterns come from real post-level behavior, not pundit opinion about the category. It means the concept slate is grounded in mechanics that are already breaking out, which is a different document than "five ideas we came up with on a call."

And critically: Virlo metadata is labeled metadata-only until video intelligence actually runs on the video asset. The agent will not infer visual beats from a caption. If you see a claim about a visual mechanic, it's because the video got analyzed, not hallucinated.

The kit ships with 7 research modes. Each mode has a different connector requirement and produces a different confidence label. The mode label is never faked.

If a connector is missing, the mode label degrades honestly. It doesn't silently pretend. quick-pass means Hypothesis only. outlier-pass means Directional evidence. deep-pass with video intelligence running means Strong directional. Everything the brief claims is tagged at the confidence level the evidence actually supports.

This is the opposite of how every other strategy AI works. Most tools are incentivized to sound certain. This one is incentivized to be honest.

The non-negotiables baked into every stage. Literally the hard rules:

If you've ever had a creator strategy doc get roasted because a client caught one hallucinated creator claim, these are the rules that stop that from happening.

Section 7 of the brief is the section that changes how clients react to the doc.

It's a list of the 5 generic plays the strategy is explicitly rejecting. Before any concept gets recommended, the brief commits to what it will not do.

Things like:

When a client reads that section, they stop treating the brief like a deck and start treating it like a plan. The rejection is what makes the rest of the document feel real.

This is the tenth kit I've open-sourced in 10 weeks (ok that’s kinda cool, just counted and realized that). Ten weeks, ten kits, all free. And the recent ones are designed to fit together as a loop: the Meta Ads Kit runs your paid distribution, the Landing Page Factory builds angle-matched pages, and the Outcome Kit tells you which angles are actually producing buyers.

Creator Breakout Kit is the layer that sits in front of all three and decides what to test.

Creator Breakout Kit is the strategy layer before the paid stack. It decides what angles are worth building pages, creative, and test budget around. Then the other kits take it from there.

Before this kit existed, the loop started at "pick some ads to run." Which meant the strategy was retroactive. You'd find out what worked after you'd already burned the test budget on wrong angles.

This one moves the thinking to the front. You decide what to test before you source, before you produce, before you spin up creative. Then every downstream kit is pointed at the right thing.

quick-pass mode runs with zero connectors. Zero. You still get a brief, just with Hypothesis-only confidence labels instead of Strong directional. Add Firecrawl and Virlo when you want outlier-pass for client-facing work. Add video intelligence when the prospect is high-value enough to justify the paid strategy asset.

MIT license. Fork it. Use it on client work. Use it on your own brand. I'm not charging for the strategy layer.

Clone the repo, run preflight, configure your sources. Setup takes an afternoon.

GET CREATOR BREAKOUT KIT (FREE)

What you need:

The workflow:

Step 1: Clone the repo and drop in your API keys.

Standard stuff. Clone the repo, copy .env.example to .env, fill in your Firecrawl and Virlo keys (or leave them empty if you want to start in quick-pass mode). Run the preflight check and it'll tell you which research modes are available based on what you've filled in.

Step 2: Tell the agent about your brand.

Open Claude Code (or OpenClaw) in the repo and just talk to it:

The agent seeds the run directory, loads brand context, and asks clarifying questions if anything's missing. Two minutes of conversation per new brand. No CLI flags to memorize.

Step 3: Ask for the outlier research.

Firecrawl pulls brand-level source material from the site. Virlo pulls post-level outliers in your category. All evidence gets attached to the run folder as raw artifacts, not inferred from vibes.

Step 4: Ask for the brief.

The agent runs enrichment, builds the concept slate with the four framework labels, scores every concept by strategic fit and evidence strength, and renders both the markdown version and the dark-terminal HTML artifact. Open the HTML in a browser. That's the show-off version you can send to a partner or paste into a Notion.

Step 5 (optional): Upgrade the evidence.

If the prospect is high-value enough to justify the paid strategy asset, just tell the agent:

It pulls the actual video assets for the strongest outlier posts, analyzes them with Gemini, and re-synthesizes the brief. Confidence labels upgrade from "metadata-only" to "Strong directional." Same brief, much sharper evidence.

You can also ask it pointed questions at any stage:

Not for you if you have a senior creator strategist on payroll who is already running post-level outlier analysis, mapping buyer tension to hook structures, and rejecting generic plays in writing. You don't need this. You are the 1%.

Also not for you if you want the agent to source the creators for you. It does not. That's deliberate. Creator names in the brief are comps only. Sourcing is a separate step, and it should be, because the strategy layer and the sourcing layer are two different jobs.

V1 is TikTok + Reels focused. The breakout pattern research pulls from short-form platforms. YouTube long-form support is coming but isn't native yet.

Virlo is the paid dependency. The recommended default mode (outlier-pass) requires Virlo (you can now pay for API use). The kit will run without it in quick-pass or brand-pass mode, but the concept slate will be labeled Hypothesis-only. Decide what the output needs before you pay.

Video intelligence is opt-in. Upgrading evidence from metadata-only to Strong directional means actually pulling and analyzing video assets. That has a cost per run. I wired it up so you decide per prospect whether the strategy asset is worth the upgrade.

The brief is a test plan, not production. It tells you what to run first, alongside, and backlog. It does not write the actual creator brief for sourcing. That's the next step, and I wrote a separate sales-handoff skill that helps with follow-up copy but does not replace a real production brief.

Creator names stay comps. Always. The kit will never tell you a creator is booked, available, or committed. That's a guardrail, not a limitation, but it's worth calling out because it's different from tools that imply a roster.

Full system free on GitHub:

You get:

No upsell. Fork it.

Stop sourcing creators before you strategize concepts. Start from what's actually breaking out.

Go big,
Matt

P.S. If you missed the recent OpenClaw kits, they're all still free and they plug into each other: Outcome Kit, Landing Page Factory, ScrollClaw, Meta Ads Kit, SEO Kit. This one is the strategy layer that runs before any of them. It decides what angles are worth the other kits' attention in the first place.

We’ve been running these agents for founders. Reply to this email and say ‘agent’ if you want on the list.

🎁 Get rewarded for sharing! My team grew NEW accounts to over 50 million views in just a few months. We made an AI viral hook generator so you can follow the same hook frameworks that we do.

Invite your friends to join, and when they sign up, you’ll unlock our AI Viral Hook Generator—the ultimate tool for creating scroll-stopping content.

{{rp_personalized_text}}

Copy and paste this link: {{rp_refer_url}}