AI Policy

Thinqr's product is built around AI. We have strong opinions about how that should be done responsibly. This page is our public commitment.

We do not train on your data

Customer code, candidate code, interview transcripts, scoring rubrics, and any other data generated on the Thinqr platform are never used to train, fine-tune, or otherwise improve AI models — ours or anyone else's. Not by us, and not by our AI sub-processors.

We use enterprise tiers of Anthropic and OpenAI that contractually disable training on input data, with zero-retention logging.

AI is encouraged in interviews

Unlike legacy interview platforms, we encourage candidates to use AI tools — Copilot, Claude, ChatGPT, Cursor, anything. Pretending otherwise is dishonest about how modern engineers actually work. We measure how well candidates use AI as a skill, not as a violation.

The orchestrator is explainable

When the orchestrator makes a judgment or assigns a score, the evidence is captured and auditable. Reviewers can see exactly what the AI saw, what it concluded, and why. You are never asked to trust a black box.

Humans make the final call

Thinqr does not auto-reject candidates. Scores, signals, and flagged moments exist to help your team decide — not to replace the decision. Final hiring calls are always human.

Fairness and bias

We evaluate our scoring models for disparate impact across demographic groups and publish our methodology on request. If you see bias, report it: fairness@thinqr.io.

Incidents and model changes

Material changes to the AI models or policies powering Thinqr will be communicated in advance on our changelog and by email to customer admins.