For attorneys who talk for a living.

Where Bonfiyah fits in legal practice

Litigators, criminal defense attorneys, and family-law practitioners spend most of their day in conversations that need to be remembered with precision. The tools that exist either capture the audio without doing anything with it (Voice Memos, JPR) or capture conversations on Zoom while ignoring the in-person work that constitutes most of the practice (Otter, Notta).

Bonfiyah is built for the in-person work — client intake at the office, witness prep in the conference room, the courthouse-steps conversation with opposing counsel — with the consent management and audit-trail features the practice requires.

Client intake

Bonfiyah's consent module captures the client's verbal "yes" at the start of intake, attributes it to their speaker fingerprint, and timestamps it alongside the recording. The Story Mode recap (Pro AI) gives you a narrative version of the intake to review later — the order of disclosures, the moments of hesitation, what the client emphasized vs. what they downplayed.

The transcript supports search across the firm's entire intake history. "Find the client who mentioned previous counsel had threatened sanctions" becomes a 5-second query.

Witness prep

Truth Layer is the feature designed for this work. Across multiple prep sessions with the same witness, Bonfiyah flags inconsistencies — places where the witness's account has shifted, where dates have changed, where a third-party detail has been added or dropped. These aren't always problems, but they're always worth knowing about before opposing counsel finds them in deposition.

Pre-Brief generates a one-page summary before each prep session — what was covered last time, what's still unresolved, suggested questions to bring.

Deposition review

After deposition, drop the official transcript (or the audio if you're working from your own recording) into Bonfiyah's import flow. Truth Layer cross-references the deposition against the witness's prior recorded statements; Promise Tracker captures every commitment opposing counsel made on the record.

Settlement and negotiation

Settlement conversations live and die by precision about who agreed to what. Pro AI's action-item extraction with owner + date + source-quote is the level of fidelity that lets you write the settlement memo from the recording instead of from your notes.

Audit trail and defensibility

Every recording in Bonfiyah carries: a SHA-256 hash of the audio file, a timestamp at start, a location stamp (optional), the consent log with each speaker's verbal opt-in, the chain of any redactions or edits. Exports are PDF with the hash inline, suitable for chain-of-custody arguments.

This is not a replacement for a court-certified transcript. It is a defensible record of the conversation as it was had, generated by a tool you control end-to-end. Talk to your tech-competence CLE provider about your jurisdiction's specific evidentiary standards.

Attorney-Client Privilege markings that carry through

Bonfiyah is the only Apple-native recorder with explicit Attorney-Client Privilege markings built into the workflow. (The same privilege picker also offers a Doctor-Patient variant for medical practitioners — see the medical use case.) Two levels of granularity:

Per-speaker. Tap a speaker's shield in the playback view and mark only their words as privileged. Useful in multi-party meetings where one attorney is present alongside non-privileged participants. The speaker's badge turns red with a lock-shield icon; their pill in every email summary and every PDF export reads ATTORNEY-CLIENT PRIVILEGED instead of the consent state.

Per-story. Flip the "Mark Attorney-Client Privileged" checkbox in the story metadata row. The whole conversation gets the privilege treatment: a red 🔒 ATTORNEY-CLIENT PRIVILEGE banner at the top of every email summary, the same banner at the top of every PDF export, a [ATTORNEY-CLIENT PRIVILEGE] prefix on outgoing email subjects, a red pill on the Stories list cell, and a plain-text banner on SMS / Markdown share fallbacks.

Toggling privilege is logged in the same audit trail as consent grants and revokes — chain-of-custody hash, timestamp, method. The export PDF reflects whatever the state is at export time.

Privilege is independent of consent. A speaker who is both Granted and Privileged renders as Privileged in all surfaces — privilege is the more specific signal for handling. Toggle privilege off and the underlying consent state returns.

Read the full consent + privilege reference →

Privilege and confidentiality

Live captioning runs on-device; full transcription and AI processing run on Bonfiyah's backend under no-training commercial contracts (raw audio deleted within 7 days, nothing used to train any model). For privileged communications, evaluate whether server-side processing under those contracts meets your firm's policy — running with iCloud sync off keeps recordings off iCloud, but the transcription and AI passes still use our backend; if your policy requires fully local processing, Bonfiyah is not the right tool for those matters. iCloud sync, if you enable it, encrypts in transit and at rest under Apple's standard mechanisms; you should evaluate whether your firm's data-security policy considers this sufficient for privileged communications. If not, run Bonfiyah with iCloud sync off.

No transcripts are used to train any AI model. Ever. This is in our privacy policy as a binding commitment.