<aside> 🚀

🚀 FREE STRATEGY CALL

Want to grow your product through viral reach?

We help teams across iGaming, mobile apps, AI products, and SaaS build creator-led organic distribution generating tens of millions of views at CPMs starting from $0.03.

<aside> 🚀

Book a free strategy call

</aside>

On the call you'll receive:

Eight agents, nine scoring frameworks and 30 ready to use prompts that turn eight streams of raw input into a ranked weekly content plan a team can film on Monday.

What is inside

Most content teams do not run out of ideas. They run out of a way to choose between them. The AI Creative Director is the weekly system that makes the choice. Eight agents read eight separate streams of input. Nine scoring frameworks turn that reading into a ranked list. The output is ten concepts, five hooks for each one, a named creator, a brief and a testing hypothesis.

Nothing on this page replaces a human editor. The system removes the weekly argument about what to make next. It does not remove the person who approves it. Organic distribution built this way complements paid media and lowers blended acquisition cost across the account. Every agent below returns a draft. Every draft has a named owner who signs it off.

Layer Count Where it lives
Input agents 8 The eight input agents
Diagnosis procedure 1 The weekly creative diagnosis
Scoring frameworks 9 The nine scoring frameworks
Concept and brief system 1 From ten concepts to briefs
Production board 1 The weekly production board
Approval checkpoints 7 Approval checkpoints and handoffs
Ready to use prompts 30 The prompt library

How to run any agent on this page

Two ways. Same system prompt.

  1. Fastest, inside Claude. Open a Claude Project or a new chat, paste the agent's system prompt, then paste your inputs. You get the output straight away.
  2. As an automation, inside Relevance AI. Create a new agent, set the model to Anthropic Claude, and paste the system prompt as the core prompt. Wire in your inputs and outputs.

Start with method one to test an agent. Move to method two once you want it running on repeat.

<aside> ⚠️

Every platform treats paid partnership disclosure differently, and the rules change without notice. Nothing this system produces goes live without a named human approving the final cut, the caption and the disclosure wording. Agents draft. People approve. Confirm creator licensing terms in writing before any clip runs beyond the creator's own account.

</aside>

The eight input agents

Eight inputs. Eight agents. One per input, so a weak stream never hides inside a strong one. Run them in order, on the same day each week, into one document. Each returns a short sourced note, and those notes are the only evidence the diagnosis may use.

Keep the notes factual. Scoring happens later, against concepts. An input nobody reads is not an input.