We thought the maths was wrong.

The mission

Strip out the hours of paid human reading that today's litigation forces on its clients, without giving up the accuracy, the chain of custody or the court-acceptability of the resulting evidence.

Family disputes are uniquely evidence-heavy, and increasingly they don't stay purely civil: a C100 application can trigger cross-allegations that pull a matter toward criminal territory. The same disputes that once lived in a few letters and a phone bill now sprawl across years of chat, photos, voice notes and shared calendars. The body of evidence has exploded; the way it's reviewed has not.

We're built for the place where the maths is most broken: people paying £5,000–£25,000 in solicitor reading-time fees, when the same organising job can be done in software for a fraction of that. We start with family matters and extend to the wider civil and criminal-adjacent disputes that grow out of them.

The values

• Quote-grounded or it didn't happen. Every claim our software makes is footnoted to a source line. No "the AI says so".
• You can always export and leave. Lock-in is not part of the model. Your case data is yours.
• Privacy is a load-bearing feature. Not a marketing claim. Not training data. Encrypted, isolated, deletable with a receipt.
• We surface, we don't judge. The platform organises the record and preserves context. It never decides who is right, assigns blame, or reaches a verdict, that is the court's role.
• Honest about what AI can and can't do. We tell you where the model is confident and where it's guessing, and every figure is a count or an interval you can trace to source.