Setting up an AI Provider

The AI assistant needs an account with an AI provider and an API key from that account. Pedigree Forge supports Anthropic (the Claude models) and OpenAI (the GPT models). You can enter a key for one or both.

Tip — If you are choosing between them, choose Anthropic. At present it is noticeably more stable in Pedigree Forge and produces better results, particularly on tasks that involve reading your data and making changes to it.

Getting a key

An API key is not the same thing as a ChatGPT or Claude subscription. A consumer subscription gives you a chat website; an API key gives an application like Pedigree Forge permission to make requests on your behalf, billed by usage. A subscription will not work here, and an API key does not require one.

For Anthropic:

Create an account at console.anthropic.com.
Add a payment method and buy some credit — a small amount is plenty to start with.
Go to the API keys section, create a key, and copy it. It is shown once only, so copy it before closing the page.

For OpenAI the process is the same at platform.openai.com — create an account, add credit, then create a key under API keys.

Keep the key private. Anyone who has it can spend money on your account. If a key is ever exposed, delete it in the provider’s console and create a new one; nothing else needs to change.

Entering the key

Open Tools > AI > AI Settings and paste the key into the Anthropic API Key or OpenAI API Key field. The fields are masked, so paste rather than type.

Then choose the Provider — this is the one the chat will actually use. Providers you have supplied a key for are shown in bold; the rest are marked (no key). Click OK and the choice takes effect immediately; there is no need to restart.

If you enter both keys you can switch between providers at any time by reopening this dialog. There is no automatic fallback: the provider you select is the one used.

What it costs

You are billed by the provider for the text sent and received, counted in tokens. In practice:

A simple question about your data typically costs a fraction of a penny.
A long task — reading many records, then making changes — costs more, because the assistant reads before it writes.
A long conversation costs more per message than a short one. The whole conversation is sent again with each new message, so a rambling session gets progressively more expensive. Clearing the chat when you move to a new topic keeps costs down.

After each reply, the chat panel shows the tokens used and the estimated cost for that exchange, so you are never guessing.

Both providers let you set a spending limit on your account. Setting a modest one is a sensible precaution while you find your feet.

If the chat says no provider is available

Tools > AI > AI Chat reports that no AI provider is available when the selected provider has no usable key. Check that the key was pasted in full, that it belongs to the provider selected in Provider, and that your provider account has credit — a key on an account with no credit authenticates but every request fails.