Matter file analysis
Synthesize across every document in a matter — pleadings, deposition transcripts, contracts, correspondence — without uploading any of them to a vendor cloud.
Built with a 40-year law firm
Built by lawyers.
For the firms BigLaw forgot.
Jubal is a modern AI legal partner — engineered alongside a 40-year practice we've worked with for three decades. Designed for solo and midsize firms that need a partner, not another vendor.
Founding cohort opens summer 2026 · 100 firms
Your work is never trained on. Never shared. Yours, not ours.
- No vendor vector database
- Never trained on your work
- Never shared with anyone
The firms legaltech overlooked
Most legal AI was built for the AmLaw 100. Jubal wasn't.
There are about 100,000 attorneys in the AmLaw 100. There are 1.2 million attorneys outside it. The well-known legal AI tools chose the first group. We chose the second.
Jubal is built for solo and midsize firms — the practices that handle most of the law in the country and the least of the budget for tools.
AmLaw 100100,000
Everyone else1,200,000
What Jubal does
Four jobs Jubal handles for your firm.
Deposition and discovery prep
Surface key facts and timeline inconsistencies across thousands of pages. The work that took your associate a week, in an hour.
Drafting with your firm's voice
Draft in the language your firm already uses — pulling from your past briefs, your clauses, your precedent. Not generic AI prose.
Your firm's knowledge base
Search every matter your firm has ever worked. The senior partner's knowledge made addressable for everyone in the firm.
How it works
Three steps. Then it's part of how your firm works.
Point Jubal at a matter folder.
Jubal works on the documents you already have. No data migration, no ingest pipeline.
Jubal indexes everything privately.
Each attorney builds a private index on their own machine — full text, structure, cross-references. For multi-attorney matters, secure firm-scoped sync keeps your firm in sync across offices.
Ask, draft, review.
Ask questions across the matter. Draft new documents in your firm's voice. Review contracts against your playbook. With citations to where every claim came from.
The trust pillar
Your client work is yours.
Most legal AI products take ownership of your documents the moment you upload them. They chunk them into a vendor-hosted vector database, feed them to a third-party model, and reserve the right to “improve their service” with your data. Jubal is engineered the other direction.
We never train on it.
Your matter files, your queries, the documents Jubal generates for you — none of it is used to improve any model. Not Jubal's. Not any of our model providers'. Not anyone's. This is a contractual posture with every provider Jubal uses, not a marketing claim.
We never share it.
Not with our partner firm. Not with other firms. Not with researchers. Not with anyone. Your client work belongs to you and your clients.
Your matters stay yours.
Matter content lives on your firm's machines. When multi-attorney work requires sync across offices, content moves only between seats your firm authorizes — encrypted in transit and at rest. We have no use for your matters beyond answering the prompt you sent. We don't keep them for training, evaluation, or any other purpose.
- No vendor vector database
- Best frontier models — no training posture
- Per-matter walls and audit logs
- Encrypted at rest and in transit
Built with practitioners
Built with BocaLaw.com.
Jubal was built by software engineers and lawyers, together. BocaLaw.com has practiced for 40 years and has worked with us for the last 30 — long enough to know what every other legal tool gets wrong. Their attorneys are co-designers, not advisors. They sit in the same standups. They use what we ship before anyone else does.
And to be explicit: they do not see your client work. They cannot. The architecture doesn't allow it, and our agreements forbid it. Their value to Jubal is craft and judgment — not access to your matters.
When you join the founding cohort, that's the bench you're getting access to.
“We've been waiting thirty years for someone to build this. We finally stopped waiting and built it ourselves.”
Common questions
Common questions.
Does Jubal train on my client work?
No. Your matter files, your queries, and the documents Jubal generates for you are not used to improve any model — Jubal's, our model providers', or anyone else's. This is contractual with every provider Jubal uses, not aspirational.
Does your partner firm see my work?
No. Our partner firm helped design and build Jubal. They have no access to your matter documents, your queries, or any other firm's data. The architecture doesn't allow it and our agreements forbid it.
Where do my files actually live?
Each attorney indexes their matters on their own machine — no vendor vector database, no cloud upload of your raw documents. For multi-attorney matters, Jubal can sync securely across firm seats; access stays scoped to the people your firm authorizes, and we never train on your work.
When does the beta open?
Founding cohort opens summer 2026. We're inviting 100 firms first; the broader beta opens after the cohort has had a few weeks to push on the product.
What models does Jubal use?
The latest tested frontier models from every major provider — Anthropic, OpenAI, Google. Jubal automatically picks the best model for each task and combines providers when that gives a stronger result. All under no-training posture. You don't pick the model; Jubal picks the right one for the work.
What does it cost?
Founding cohort pricing is available to firms who join the waitlist. Public pricing publishes at general availability.
Be first when the founding cohort opens.
Six quick questions on the next page get your firm priority for the founding cohort.
Founding cohort opens summer 2026 · 100 firms
Your work is never trained on. Never shared. Yours, not ours.