How a question becomes an answer

  1. You ask a question on the homepage or from inside your account.
  2. We strip obvious personal data and search the UK corpus — statutes, case law, our solicitor-reviewed articles.
  3. We retrieve the most relevant passages from articles a UK solicitor has signed off.
  4. Our AI drafts an answer using only those passages, with citations.
  5. A second AI call checks the draft against the source passages. If the answer doesn’t hold up, we refuse.
  6. The answer reaches you with the citations and a confidence label.
  7. Everything is audit-logged so any answer can be replayed for any audit.

What we check before an answer ships

  • Grounded in our library. Every paid answer cites at least one passage from an article reviewed by a panel UK solicitor.
  • Verified before you see it. A second pass reads the answer alongside the source and refuses if they disagree.
  • UK employment law only. Questions outside our scope (criminal, immigration, family, tax) are refused with a clear reason.
  • Sources are clickable. Every citation links to the article or statute it rests on.
  • We say how confident we are. High, medium, low or refused — calibrated against the verification pass.
  • Audit trail. Every interaction is logged with what was asked, what was retrieved, what was returned.
  • Solicitor sampling. A panel UK solicitor reviews a 5% sample of paid answers each week. Their feedback flows back into the library.

Content review

Articles in our library are drafted, reviewed by a panel UK solicitor, and only then marked published. Articles older than twelve months are flagged for re-review automatically. The list of who reviewed what is published in the Trust Centre.

When AI is not enough

AI answers are good at general questions. For a specific situation that affects your employment — a settlement offer, an active disciplinary, a discrimination claim — book a 30-minute consultation with a panel UK solicitor. The AI says so explicitly when it thinks the question warrants it.

What we never do

  • Train any model on the text of your questions.
  • Sell your data, share it with employers or feed it into ad networks.
  • Generate citations our search didn’t actually return — every citation is to a real passage we hold.
  • Skip any of the checks above to ship faster.

Sub-processors, hosting, retention and your rights under UK GDPR are in the Trust Centre.