Free Tool · Runs on your device
The Bar-Safe AI Check
Ten questions about your current or proposed AI intake vendor, scored against the Supervised Intake Standard. Nothing is sent anywhere. Your score updates as you answer.
- 01
Does the caller hear it is AI in the first sentence?
- 02
When a caller asks “can I sue,” what does it do?
- 03
Can it state a fee or say “we’ll take your case”?
- 04
Do you have the vendor’s answer on what happens to a declined prospect’s information?
- 05
Who decides conflicts?
- 06
Are recordings/transcripts used to train the vendor’s model?
- 07
Can you name who at the vendor can read your transcripts?
- 08
Did a lawyer at your firm approve the script before go-live?
- 09
Can a caller reach a human by asking once?
- 10
Does the script change by state (recording consent, disclosure wording)?
0 of 10 answered
Send this to any vendor.
Seven questions, copy-ready — the same seven the Standard scores against. ShiFt’s own answers are on the Supervised Intake page.
Copy-ready
We are reviewing AI intake vendors against the Supervised Intake Standard. Please answer these seven questions in writing.
- 1.Does your AI tell the caller it is AI in the first sentence of every call, text, and chat?
- 2.When a caller asks for legal advice, does it redirect to an attorney — and is that redirect scripted and logged?
- 3.Can it state a fee, predict a result, or say the firm will take the case? If so, how is that turned off?
- 4.What happens to a declined prospect’s information — is it captured minimally, flagged, retained per our rule, and deletable? Confirm in writing.
- 5.Who runs the conflict check — your AI, or a person at our firm?
- 6.Where are recordings and transcripts stored, who can read them, and are they ever used to train models?
- 7.Can you show a lawyer approved the script, with a sign-off log, version history, and an export on request?
This tool runs on your device and shows its work. Your answers are not sent anywhere. It summarizes public guidance for planning and is not legal advice. Claim disclosures · Methodology