AI Disclosure.

Creator Agent OS uses AI to summarize brand emails, score rates, analyze contracts, and draft responses. This page explains which features use AI, which provider does the processing, what happens to your data, and where the limits of AI-generated output sit.

1. Which features use AI

The following features process your content through an AI model:

• Inbound email summarization. When you forward a brand email to your unique inbound address, the email body is sent to the AI to extract the brand name, fee, deliverables, timeline, usage rights, and a one-sentence editorial summary.
• Rate context and analysis.The rate calculator's narrative output and the deal-specific rate-analysis chip use AI to translate numeric ranges into plain-English context (“your ask is in the upper third of comparable creator deals” etc.).
• Contract analysis. When you upload a contract PDF or DOCX, the extracted text is sent to the AI to flag risky clauses (exclusivity, indemnity, usage rights, payment terms) and propose negotiation language.
• Response drafting. The draft-a-response feature sends the inbound message plus your profile context to the AI to generate a first-pass reply in editorial voice.
• DM/screenshot parsing. When you upload a screenshot of a brand DM, the image is sent to a vision model to extract the message text, then to the AI for the same summarization as email.

2. Which AI provider

All AI processing in Creator Agent OS is performed by Anthropic's Claude API (anthropic.com). We do not use OpenAI, Google, or any other AI provider for the features listed above.

3. How brand-email content is handled

When you forward a brand email, the message body (and subject line) is transmitted over TLS to Anthropic's API for parsing. The parsed result returns to our servers and is stored in your account's database row as the AI summary you see in the dashboard.

Anthropic does not use commercial API traffic to train their models. That includes everything Creator Agent OS sends to them. See Anthropic's commercial terms for the authoritative statement: anthropic.com/legal/commercial-terms.

4. How contract content is handled

Uploaded contracts are stored in private Vercel Blob storage on our infrastructure. The extracted text (PDF/DOCX → plain text) is transmitted over TLS to Anthropic's API for risk analysis. The analysis returns to our servers and is stored against the relevant deal row. Same training opt-out as above — Anthropic does not use the contract text to train models.

5. AI outputs are not professional advice

Rate calculations are estimates, not financial valuations. The rate engine uses calibrated multipliers against public-market deal data. It produces a range, not a guarantee of what any specific brand will pay.

Contract analysis is not legal advice. The AI flags clauses that match risk patterns and proposes alternative language, but it is not licensed to practice law and may miss issues a real attorney would catch. For contracts with material financial, ownership, or liability exposure, hire a lawyer.

Response drafts are starting points, not final messages. Always review and edit AI-generated copy before sending it. The AI doesn't know your full negotiation history, your priorities, or the tone you'd use with that specific brand.

You — the creator — make every final decision. AI outputs are inputs to your judgment, not substitutes for it.

6. AI accuracy and limitations

AI models occasionally produce wrong, incomplete, or fabricated outputs (“hallucinations”). The rate engine occasionally misreads a non-standard deal shape. The contract analyzer occasionally flags benign clauses or misses subtle ones. The response drafter occasionally suggests phrasing that doesn't match your voice.

Treat AI outputs as a fast first pass — useful for triage, useful for getting unstuck, not authoritative. When you spot an AI mistake that matters, the in-product feedback widget routes it directly to us.

7. Data retention by Anthropic

Anthropic retains API traffic per their commercial data retention policy. As of this writing, Anthropic automatically deletes commercial API inputs and outputs within 30 days, unless a longer retention period is required for trust-and-safety review. The authoritative and current version of their retention policy: privacy.anthropic.com — data retention.

We do not separately retain data on Anthropic's infrastructure beyond what their retention policy allows. The AI outputs we store in our own database (your dashboard's AI summaries, contract analyses, etc.) are governed by our Privacy Policy.

8. Opting out of AI features

AI processing is core to Creator Agent OS. The inbound-email summary, the rate context, the contract analysis, and the response drafter all depend on it — without AI, the product is a static spreadsheet.

If you don't want your data sent to Anthropic, you can:

• Not forward brand emails to your inbound address (manual deal logging still works)
• Not upload contracts (manual deal logging still works)
• Not use the calculator wizard's narrative output (the calculator's numeric range is computed by deterministic engine code, not by AI)
• Not use the response drafter

In practice this leaves a much thinner product. If that fits your needs, the manual-only flows are fully supported. If not, opting out means not using Creator Agent OS.

9. Questions

Anything unclear about how AI is used here, email hello@creatoragentos.com.

Studio 1111 LLC
New York, NY