Secrets (AI Keys)

The Secrets tab is where you add your own Anthropic API key to this product. This is a Pro plan feature that lets you route core Claude model usage through your own Anthropic account rather than consuming Productised model credits.

Claude Runtime

Productised runs all AI Products on a fixed Claude runtime for consistent output quality. The provider (Anthropic) and runtime model (Claude Sonnet) are determined by Productised and cannot be changed. This ensures that chat quality, report generation, and document output remain consistent across all products on the platform.

Adding your Anthropic API key

If you are on the Pro plan, you can add your own Anthropic API key here. When a key is set, core Claude model usage for this product is charged to your Anthropic account rather than drawing from your Productised model credits.

To get an Anthropic API key, visit console.anthropic.com/settings/keys.

Enter your key in the Anthropic API Key field (formatted as sk-ant-...) and click Save.

What your key covers

When your Anthropic API key is set:

• Core AI Product runs — chat, collection, follow-up advisory, visual results, and document generation — use your Anthropic account
• Core model usage is billed directly by Anthropic on your account at their standard rates
• Claude output quality and model selection remain fixed by Productised

What your key does not cover:

• Knowledge base retrieval — this uses Productised managed infrastructure
• Research / web search capabilities — this uses Productised managed infrastructure
• Image generation (getIMG) — this requires a separate getIMG API key

Your Anthropic account's spending limits, rate limits, and billing rules still apply.

Legacy provider notice

If you previously configured this product with a different AI provider, you will see a notice asking you to add a new Anthropic API key before saving. Adding a key migrates the product fully onto the Claude runtime.

Credit usage estimator

The Secrets tab includes an optional usage estimator. Select your product's approximate question load, conversation depth, and result type to see whether your expected usage is Low, Medium, High, or Very High. This helps you understand the relative cost profile of the product before deploying it to clients.