Dictation for insurance field adjusters

The idea

Independent adjusters get paid per closed claim, and the bottleneck is the write-up: hours every evening converting voice memos and photo dumps into the narrative format each carrier demands. The market has already blessed dictation as a category people pay for — generic dictation tools charge $8-15/mo and got enthusiastic verdicts — but they output raw text, which still leaves the adjuster formatting at midnight.

The wedge is the carrier template library, not the transcription. Speak the walkthrough once, get back a structured narrative in the carrier’s expected format with a captioned photo log, timestamps aligned so each caption describes what the adjuster was actually looking at. Transcription is a commodity API call; knowing what a claim narrative for a wind/hail loss looks like is the product, and it justifies $40-60/mo against tools charging $10 for words on a page.

Who pays

The buyer is the independent adjuster or small IA firm owner, at $49/mo solo and $39/seat for firms — recovered in the first claim closed faster. During CAT season adjusters run 10+ claims a day and will pay anything that shortens evenings. Distribute through adjuster Facebook groups, CAT deployment forums, IA firm owners who standardize tools for their roster, and adjuster licensing course communities.

MVP scope

• Offline-first recording with chunked upload and a photo camera in the claim flow
• Whisper transcription plus Claude narrative generation into structured sections
• Five carrier-style narrative templates with per-carrier profiles
• Photo captions auto-matched to transcript timestamps, editable
• PDF and DOCX export
• Skip for v1: Xactimate integration, sketch/measurement tools, firm dashboards, direct carrier submission

Build ClaimVoice, a mobile dictation app for independent insurance adjusters. Stack: React Native with Expo, Whisper API for transcription, Claude API for narrative generation, Supabase for auth, Postgres, and photo storage.

Core flow: adjuster opens a claim, taps record, and narrates the walkthrough room by room while taking photos inside the app. Audio records in chunks and uploads opportunistically — the app must work offline in basements and rural sites, queueing everything until connectivity returns. Back at the truck, the adjuster taps Generate: Whisper transcribes, then Claude produces a carrier-formatted claim narrative (risk description, cause of loss, damage observations by area, scope recommendations, subrogation notes) plus a caption for each photo matched by timestamp to what was being said when it was taken.

Screens: claim list, active claim with record button and camera, photo log with editable captions, narrative review editor, template picker, export screen (PDF and DOCX). Templates are the moat: ship five carrier-style narrative formats and let users tweak section headers and ordering per carrier profile.

Non-obvious details: timestamp-align photos to the transcript so captions describe the actual subject; store an editable structured intermediate (JSON of sections) rather than raw prose so template switching re-renders without re-calling Claude.

Pricing: $49/mo solo, $39/seat for firms of 3+, 14-day trial, Stripe billing via web portal.