Question 1

Can Ana choose the OpenAI model?

Accepted Answer

Yes. Ana can select any model available to your OpenAI organization — GPT-4, GPT-4o, GPT-3.5, embeddings, or moderation — based on the question and cost constraints. Admins can restrict which models are allowed to control spend and compliance.

Question 2

Does this send warehouse data to OpenAI?

Accepted Answer

Only the data you explicitly pass into the OpenAI call. Ana is transparent about what data leaves the sandbox. Admins can configure allowlists that prevent certain columns or tables from being passed to external APIs, giving you strong control over data egress.

Question 3

How is the OpenAI key secured?

Accepted Answer

The API key is encrypted at rest in TextQL's credential vault and injected by our egress proxy at the network layer. Ana's Python sandbox never sees the raw key value. You can rotate the key at any time from the TextQL admin console without changing prompts.

Question 4

Are OpenAI costs tracked?

Accepted Answer

Yes. Every OpenAI call Ana makes is logged with the model, token counts, and approximate cost. Admins can set per-org or per-user spend caps and receive alerts when thresholds are approached, preventing runaway costs from analysis workflows.

Question 5

Can I use Azure OpenAI instead?

Accepted Answer

Yes. If you use Azure OpenAI for compliance or data residency reasons, an admin can configure the connector to point at your Azure OpenAI endpoint and whitelist that domain. The same sandbox security and credential handling apply.