Chat Forensics — Upload a WhatsApp export and get a plain-English report that quotes its evidence.
Upload a WhatsApp chat export and get a plain-English AI report: timeline, patterns and risks, with every finding quoting the messages it rests on. Names are replaced before the AI reads the chat.
The problem
Long, messy chat exports with voice notes, images and video; readers who need findings they can check rather than an AI's opinion; and personal data that should never reach a model unredacted.
What we built
An ingestion pipeline for chat text and media, name redaction before analysis, reports whose every finding cites the source messages and states its limits, a free preview, and pay-per-chat checkout with automatic refunds if a report fails.
What it does today
A timeline, patterns and risks from a whole chat history in minutes, every finding linked to the messages behind it, with names removed before the AI reads anything.
— Under the bonnet
What makes a product like this hard to build
- 01
Real chat exports are messy
WhatsApp exports vary by phone, language and date format, and include voice notes, images and video. Turning them into a clean timeline is the first hard problem.
- 02
Privacy before analysis
Names are replaced before the AI reads anything, which means finding and consistently replacing every person in the conversation, including nicknames.
- 03
Findings that quote their evidence
Every point in the report cites the messages it rests on and states its limits, so the reader can check it rather than take the AI's word for it.
- 04
Paying per report
A free preview, pay-per-chat checkout and automatic refunds when a report fails all have to be reliable, because each report is a one-off purchase.
— Estimated build cost
What a build like this costs
£230,000–£340,000
Conversation-analysis report product · indicative range for a comparable build
Sized from the codebase we built: 296k lines of production code, 24k lines of tests, 113 database tables, 216 API routes.
A product like Chat Forensics is a bespoke build, not a template: most of the cost is in the hard parts above, plus testing against real data before anyone relies on it. Where your version lands in the range depends on the factors listed here. You get a fixed price once scope is agreed — usually after a short discovery sprint. Hosting and AI model usage are billed at cost on your own accounts.
Get an estimate for yoursWhat moves the price
- Export formats and media types supported
- Report depth and length
- Redaction accuracy requirements
- Payments, previews and refunds
- Processing cost per chat
Built with
- React
- TypeScript
- Express
- PostgreSQL
- Drizzle ORM
- OpenAI
- Stripe
- Python workers
Designed, built and run by our Manchester team. Want something similar? See our AI development services, AI automation and pricing.
