10-lesson course · about 12 hours
Build proactive AI agents that keep your matter record current
For Attorneys, paralegals and in-house legal teams
Everything for Legal teamsSet up and supervise bounded agents for intake, chronology, drafting and evidence that process approved new documents, preserve the matter record and bring exceptions to an attorney.
What you’ll be able to do
By the end, you can…
How it works
A tutor that meets you where you are
- Lessons that adapt to your levelTen lessons in five modules follow the matter workflow from intake to recovery testing. Explanations and examples adjust to what you already know about matter files and agents.
- An AI tutor you can ask anythingAsk questions typed or out loud while you work. The tutor helps you draft agent briefs, check exception rules and review run reports, the way a trainer would in a live class.
- Quizzes and flashcards for each lessonEach lesson has its own quiz and flashcards, so you can check your grasp of terms such as continuing brief, accepted state and verification queue.
- Progress tracking and a certificateWork at your own pace and see where you are in each module. You receive a certificate of completion when you finish.
Curriculum
The curriculum
The course follows a sample matter across five modules, one per agent, and ends with a test of the full recurring workflow. Each lesson produces a working piece of the matter record that an attorney can inspect.
Module 1Keep incoming documents organized2 lessons
1Define the proactive intake agent’s job
- Write a job statement for a named intake agent that says what it must and must not produce
- Choose between OpenAI Dots and Claude Duet for the matter workflow using criteria you can check against your organization's approvals
- Choose between a single run and a continuing brief, and justify the choice for the sample matter
- Specify approved connections, agreed reports, human review points, exception rules and pause and retire conditions
2Process a second document batch
- Run the intake agent on a second permitted batch and review the resulting document index changes
- Explain why an incremental update preserves the matter record better than restarting the task
- Test that out-of-scope or doubtful documents produce clear exceptions instead of silent additions
- Check the run report against the agreed report format and record your review
Module 2Maintain a source-linked chronology2 lessons
3Update events from new evidence
- Configure a chronology agent to take accepted inventory changes as its only trigger and input
- Build a chronology in which every event links to the source that supports it
- Record both the current account of an event and the reason and source for each change
- Explain why late evidence must reach the chronology and not only the inventory
4Handle gaps and conflicting accounts
- Identify gaps and conflicting accounts in a sample chronology
- Configure the agent to record conflicts side by side with their sources
- Write exception rules for when a gap or conflict goes to an attorney
- Test the agent with a batch that introduces a conflicting account
Module 3Prepare drafts when the record is ready2 lessons
5Build the proactive drafting agent
- Write an approved drafting request that states the draft's purpose, audience, jurisdiction and limits
- Configure the drafting agent to work only from the current accepted matter record
- Define when the record is ready enough for the agent to prepare or revise a draft
- Produce the supporting outputs that make each draft claim checkable
6Route uncertain claims for review
- Identify the kinds of draft claims that must go to attorney review
- Configure a verification queue that records each uncertain claim, its source and the reason it was flagged
- Check that cited authority is marked for verification in the appropriate source before use
- Explain how an attorney's review decision is recorded and fed back into the next revision
Module 4Keep evidence questions in view2 lessons
7Organize questions and evidence
- Define the bounds of a discovery packet: issues covered, sources used and what is excluded
- Configure the evidence-review agent to link each question to an issue and to the material that prompted it
- Build an issue register that tracks questions across runs
- Explain why the packet is a preparation aid and not a discovery decision
8Prioritize gaps for attorney review
- Set priority criteria the agent applies to gaps and open questions
- Produce an attorney review list that shows priority, issue, source and reason
- Record attorney decisions so the next run reflects them
- Test that a recurring run updates priorities when new evidence arrives
Module 5Test the recurring matter work2 lessons
9Run the proactive agents over several updates
- Plan a test of several update cycles, including late evidence and a conflicting document
- Run the agents in sequence and trace one change from intake through chronology, draft and evidence packet
- Check each cycle's reports against the agreed formats and review points
- Test how an attorney's answer in one cycle changes the next run
10Challenge the stop and restart rules
- Test an interruption mid-cycle and confirm the agent resumes from the last accepted state
- Test pause and resume rules, including who can trigger them
- Retire an agent and confirm the matter record and final reports are preserved
- Revise agent briefs to fix any weakness the tests reveal
Who it’s for
Made for you if…
Made for: Attorneys, paralegals and in-house legal teams
Paralegals who maintain document indexes and chronologies
You want new documents to update the inventory and chronology with source links and visible exceptions, without rebuilding the file each time evidence arrives.
Attorneys who supervise matter preparation
You want agents that prepare drafts and evidence questions within limits you set, while privilege, discovery decisions and the use of any draft stay with you.
In-house legal team members
You need recurring preparation work done through approved connections and agreed reports, with clear ways to pause or retire it, and with client information kept out of unapproved workspaces.
Questions
Good to know
What should I know before starting?
You should be able to frame a legal task and say what it must and must not produce. You should have worked with a matter file, including documents, sources and an inventory or chronology, and understand why source references, jurisdiction, access boundaries and open questions must stay visible. Days 1 and 2 of the Legal track, or equivalent experience organizing a matter file, is the expected starting point.
Do I need to write code?
No. The course covers configuring and supervising agents through written briefs, rules and reports. Writing code and building custom integrations are not covered.
Which AI tools does the course use?
You choose between OpenAI Dots and Claude Duet for the matter workflow. You make the choice using criteria you can check against your own organization's approvals.
Will the agents make privilege or discovery decisions?
No. Privilege determinations and discovery decisions stay with the attorney. The discovery packet you build is a preparation aid, and uncertain claims are routed to attorney review.
Does it cover legal research or court rules?
No. Substantive legal research method, case-law analysis, jurisdiction-specific court rules and procedure, and trial advocacy are outside this course. You do learn to mark cited authority for verification in the appropriate source before a draft is used.
How long does it take, and is there a live class?
Expect about 12 hours. The course is self-paced with an AI tutor, and there are no live trainer-led sessions.
What is the difference between a single run and a continuing brief?
A single run does the task once. A continuing brief keeps the agent working as approved evidence arrives. In the first lesson you choose one for the sample matter and justify the choice, and later lessons cover pause, resume and retire rules for continuing work.
