Autos: practice management and legal intelligence for law firms.
A B2B SaaS for law firms in Brazil that covers the life of a case, from intake to the filing, the deadlines and the billing, with AI as an aid to the lawyer and never a replacement.
- Each firm in its own isolated database — no shared schema between clients
- AI-assisted drafting that cites only what was retrieved — and requires human approval
- An assistant on small models, on our own infrastructure — case data does not go to third parties
- Procedural deadlines with business-day counting and a calendar
- Fees, installments and costs; reports per lawyer and per legal thesis
- WhatsApp notifications, only with consent
- Role-based access: partner, lawyer, intern, finance
- A shared base that is anonymized, with protection against re-identification
Legal work still fits poorly in the tools that exist.
We start from four hypotheses, which we want to confirm or break with firms during the pilot phase:
- Deadlines and case updates live in separate calendars and spreadsheets.
- Briefs are rewritten from loose templates, without reusing what the firm already knows.
- Generic AI does not suit confidential data: you cannot paste a case file into any chat.
- The firm's knowledge lives in people and leaves when they leave.
Why now: small models running on your own infrastructure are becoming capable enough to support a lawyer on confidential data without sending it to third parties.
AI shows up in two places, and we keep them separate.
// to build
We develop with Claude Code (and Antigravity), under specification, tests and human review. That is how a small team ships with the care of a larger one.
// inside the product
The Autos assistant runs on small models, on our own infrastructure, because case files are confidential. Today we send no client data to any frontier-model API, and we will not without clear consent and a contract.
// lines we do not cross
- Nothing generated by AI goes out without a lawyer's approval.
- Citations come only from what was retrieved from the firm's own documents.
- No client data to third-party APIs without consent and a contract.
- No metric on this page that does not exist.
What API credits would let us do: measure, with synthetic data, what the small model still cannot solve, and evaluate where a larger model fits in tasks that involve no client data.
How we work.
Security and privacy go into the design, not the end: role-based access, isolated data and auditing from the first module.
Where we come from.
- 2025-03 — the idea starts taking shape
- 2026-08 — Autos development begins
- 2026-10 — this site and the waitlist go live
- next — pilots with law firms, with no date promised
We also build custom systems, from discovery to production.
For companies that still run on paper, spreadsheets and loose messages, we deliver the whole flow:
- Management systems — web and PWA for field use, with an audit trail
- Dashboards and reports — Power BI and web dashboards over Oracle, MySQL and PostgreSQL, with AI agents helping on modelling and reports
- Integrations — Active Directory/LDAP, ERPs, APIs and messaging
- Process automation — routines, queues and flows with n8n
- Deployment — Docker, servers, databases and follow-up after launch
- Identity and interface — a system with its own visual standard, made for the people who use it
Who builds.
- Product and interface — we define the problem, design the experience and validate with users.
- Engineering and data — we turn business rules into systems that hold up.
- Infrastructure and deployment — network, servers, databases and deploy.
- Design — visual identity, colour and usability, with an outside view of the code.
A small team, pre-launch: what is here is what exists today.
Where we are and what we are looking for.
// stage, plainly
- Product — Autos, in active development
- Customers — none yet; opening a waitlist and pilot conversations
- Revenue — none yet
// how we would use the support
- Cloud credits — a staging and pilot environment, with one isolated database per firm
- API credits — model evaluation on synthetic data and development support
- Mentoring — pricing, legal (LGPD) and first pilots
// what we are looking for
- Law firms for pilots and feedback
- Mentoring and network from people who have built B2B SaaS
- Accelerator programs, cloud and AI credits
- Conversations with those who back early-stage startups
Tools.
Want to test, support or talk?
We are at the beginning and we say so plainly. If you run a law firm and want to join the Autos waitlist, have a process that still lives on paper, or follow startups and want to trade ideas, write to us.