Agents that do the work
Not a chat window. Agents read, write and organize files on your computer, several at once, and report back when they finish.
Orchard for lawyers
Orchard puts the Claude and ChatGPT plans you already have to work on your matters. Agents read the file, build the chronology, draft the brief and check the cites, then hand it to you to review.
7 days free, then $49.99/month for 6 months and $99.99/month after. Card required; cancel anytime.
Runs on your computerWalls between mattersNever used to train AIFull audit recordYouBuild a chronology from the production and the two depositions. Pin-cite every entry.
Saved to the matter folder
The gap
How it works

Point Orchard at the matter folder. Each matter becomes its own project, with its own memory, rules and record.
Type it the way you’d brief an associate. Orchard plans the work and runs Claude Code and Codex agents on it in parallel.
Drafts, logs and tables land in the folder with their sources. Anything an agent couldn’t confirm comes to you first.
What it does
Pick a job. Each example shows the brief, the agents splitting the work, the check before it reaches you, and the files left in the matter folder.
YouFirst-pass review of the new production. Tag hot documents and anything that may be privileged.
Saved to the matter folder
Examples with fictional matters. Agent output is a draft for your review.
Privacy & security
Orchard for lawyers runs on the same platform enterprises use to govern AI, so privacy isn’t an add-on. ABA Formal Opinion 512 asks lawyers using generative AI to protect client information and supervise the work. Orchard gives you the controls and the record. The judgment stays yours.

Build a chronology, pin-cite every entryHalvorsen v. Brightline onlyProduction/ (63 files)Depositions/Okafor.pdfChronology.docxChronology.docxChronology.docxOrchard’s engine runs on your machine, and agents read and write matter files right where they already live.
Orchard does not use your private workspace content to train third-party AI models.
Each matter keeps its own memory, rules and record. Recall stays inside the matter you are working in, so one client’s memory doesn’t carry into another’s work.
Work goes to the model through your own Claude or ChatGPT account, under its terms. For the most sensitive matters, switch to a local model that runs on your machine.
Orchard’s connections are encrypted in transit and sit behind access controls and monitoring.
Who asked, what each agent read and wrote, and what you approved, in an append-only record: your evidence of supervision.
For firms · Orchard Enterprise
Orchard Compliance gives partners and IT one place to see and govern every lawyer’s AI work. It presents evidence for review; it is not a certification.
What Orchard adds
Not a chat window. Agents read, write and organize files on your computer, several at once, and report back when they finish.
Facts, decisions and your corrections carry into the next session, so you never re-explain the case.
Write a rule once (your citation format, your fallback clauses, “confirm every cite against the source”) and every agent follows it.
Have one agent check another’s work before it reaches you, and decide what still needs a person.
Prompts, tool calls, files touched and approvals go into an append-only record you can review later.
Token use and cost are tracked by model, so you know what the AI spent on the work.
Pricing
$49.99/month
7 days free, then $49.99/month for 6 months and $99.99/month after. Used with your own Claude or ChatGPT plan.
Custom
Shared knowledge across lawyers, admin controls, Orchard Compliance and rollout support for the whole office.
Talk to us about your firmNo. Claude Code and Codex were built for programmers, and Orchard is how a lawyer puts them to work. You tell Orchard what you need in plain English; it runs the agents and reports back.
The ones you already have. A Claude Pro or Max plan includes Claude Code, and a ChatGPT Plus or Pro plan includes Codex. Use either one, or both.
Your files stay on your computer. The model sees what an agent sends it, through your own Claude or ChatGPT account and under that account’s terms. For matters where nothing should go to a hosted model, run the work on a local model.
Not by Orchard. Orchard does not use your private workspace content to train third-party AI models. What the model provider may do is governed by your own Claude or ChatGPT account’s terms and settings, which you control.
Yes, with Orchard Enterprise: firm-wide AI policies, roles, and Orchard Compliance to review every lawyer’s AI activity with its evidence. See enterprise.orchardai.ai.
Language models can, which is why Orchard has a second agent check the first agent’s work against its sources and flag what it can’t confirm. You still read and approve anything before it’s filed or sent.
No. Orchard is software. It drafts and organizes; you exercise the judgment and remain responsible for the work.
Seven days free. Bring your own Claude or ChatGPT plan.
Start your free trial