Skip to content

You are viewing documentation for Instruqt 2.0 Labs. For Tracks documentation, please visit docs.instruqt.com.

Using Your Own API Key


The AI assistants run on your team’s own Anthropic API key. Until a team Owner adds one, every AI feature is switched off: the lab assistant, Create with AI, and research all show a blocking message instead of starting work.

Tokens are billed to your own Anthropic organization. Instruqt spends nothing on your behalf, and AI usage is not included in your Instruqt subscription.

Only team Owners can add a key.

  1. In the sidebar, click Settings → API keys.
  2. Find the Anthropic API key section.
  3. Paste your Anthropic API key.
  4. Give the key a name.
  5. Click Save.

Saving both stores the key and makes it active. The next AI run you start uses it.

If you are not a team Owner, the Anthropic API key section is not shown. Ask a team Owner to add the key. Once the key is active, every content creator on the team can use the AI features.

Instruqt never shows the key again after you save it, and does not store the last four characters. The key is identified only by the name you give it. Choose a name you will recognize later, such as the name of the Anthropic workspace it comes from.

Instruqt validates the key with a live call to Anthropic:

  • Anthropic answers 401 — the key is wrong or revoked. Instruqt refuses it, and the key does not become active.
  • Anthropic answers with any other error, such as a timeout or a 500 — Instruqt accepts the key unchecked. A brief outage at Anthropic does not stop you configuring your account.

A saved key is therefore not a guarantee that the key works. If Anthropic was unreachable at save time, you find out on your next run.

The key is encrypted with a Google Cloud KMS key and cryptographically bound to your team, so the stored value cannot be read on any other team’s account. It is decrypted only inside the process that calls the model. It is never written to logs, traces, or run history, and no Instruqt API returns it.

Replacing and removing behave differently. This is deliberate.

ActionEffect on runs already in flightEffect on the next run
ReplaceNo effect — they finish on the key they started with.Uses the new key.
RemoveThe runs fail.AI features are switched off until a key is added.

Removing a key stops in-flight runs. The key material is deleted, so a running assistant cannot recover and the run fails.

If your key is revoked, out of credit, or rate-limited past retry, the run fails.

Rate limits come from your Anthropic organization’s tier, not from Instruqt. If runs fail on rate limits, raise the limit with Anthropic.

To recover: fix the key at Anthropic, or save a working key in Instruqt, then start the run again. Work already written to the lab is kept — nothing is rolled back.

Your own key is not automatically cheaper. Bring your own key for control over the account, the limits, and where the spend sits — not for a lower bill.

  • Anthropic only. There is no support for Vertex AI, OpenAI, or Bedrock.
  • One active key per team. There are no per-lab or per-user keys.
  • No spend figure in Instruqt. To see what your key has spent, use the Anthropic console.
  • No change history. The API keys section shows who added the active key and when. Replacements and removals are not recorded.