What is this course about?
Technology-assisted review has judicial acceptance; generative privilege screening does not yet have the same track record. This course covers where AI changes discovery economics, the validation and sampling that make a process defensible, and how to document methodology so it survives a challenge from opposing counsel.
Who is this course for?
- eDiscovery counsel and litigation support
- Litigation partners and associates
- Document review managers
- In-house litigation counsel
What do I need before starting?
- eDiscovery experience
- Familiarity with the EDRM lifecycle
What will I be able to do afterwards?
- Map where AI changes discovery economics across the EDRM lifecycle
- Apply technology-assisted review with an appropriate validation protocol
- Assess the error profile of generative privilege screening honestly
- Prove recall to an opposing party with documented sampling
- Disclose AI use appropriately in the discovery protocol
What does each module cover?
Where does AI change discovery economics?
40 minMapping the EDRM lifecycle to find where cost actually concentrates.
Objectives
- Map cost across the EDRM lifecycle
- Identify where AI changes the economics
- Set realistic expectations for savings
Topics
Activity. Map cost across a real matter's discovery lifecycle.
How does TAR work and why do courts accept it?
50 minTechnology-assisted review, the case law supporting it, and the protocols courts have approved.
Objectives
- Explain how TAR workflows operate
- Summarize the supporting case law
- Select a protocol appropriate to the matter
Topics
Activity. Design a TAR protocol for a matter with stated document volume and issues.
Can generative AI screen for privilege?
55 minThe error profile of generative privilege screening, assessed honestly against its track record.
Objectives
- Assess generative screening error rates
- Compare against traditional and TAR approaches
- Determine an appropriate role given the current record
Topics
Activity. Test generative privilege screening against a known-privileged set and measure errors.
How do you validate and prove recall?
55 minSampling and validation that produces a number you can defend to an opposing party.
Objectives
- Design a statistically valid sampling protocol
- Estimate recall with a defensible confidence interval
- Document the validation for production
Topics
Activity. Design and execute a validation protocol producing a defensible recall estimate.
How do you document for defensibility?
45 minThe record that survives a challenge, built during the process rather than after it.
Objectives
- Document methodology contemporaneously
- Record decisions and their reasoning
- Prepare for a challenge to the process
Topics
Activity. Build the defensibility record for the Module 4 validation.
What do you disclose at meet-and-confer?
45 minNegotiating the discovery protocol when AI is part of the process.
Objectives
- Determine what to disclose about methodology
- Negotiate protocol terms covering AI use
- Handle an opposing party's objections
Topics
Activity. Negotiate a discovery protocol covering AI use in a role-play.
How does proportionality apply?
40 minRule 26 proportionality and cost shifting when AI changes what is reasonably accessible.
Objectives
- Apply proportionality when AI reduces review cost
- Argue cost shifting in an AI context
- Anticipate how proportionality arguments change
Topics
Activity. Draft a proportionality argument accounting for AI-reduced review cost.
Building a defensible TAR workflow
55 minThe workshop module: a complete workflow with a validation report.
Objectives
- Assemble a complete defensible workflow
- Produce the validation report
- Prepare the methodology declaration
Topics
Activity. Complete the workflow and produce a validation report and methodology declaration.
What is the capstone project?
Defensible TAR workflow with a validation report
Design and execute a complete discovery workflow with a TAR protocol, an honest assessment of generative privilege screening, a statistically valid recall estimate, contemporaneous defensibility documentation, and a methodology declaration ready for production.
Deliverable: A validated workflow, a recall estimate with confidence interval, and a methodology declaration.
How are learners assessed?
- Recall estimate must carry a defensible confidence interval
- Privilege screening error rate measured against a known set
- Methodology declaration reviewed by a colleague acting as opposing counsel
What ships with the course?
Facilitator guide
Session-by-session running order, discussion prompts, and the questions that reliably derail a room.
Learner workbook
Exercises, checklists, and the templates each module's activity produces.
Hands-on lab environment
A sandboxed ibl.ai deployment so exercises run against real agents, not screenshots.
Assessment bank
Scenario questions and rubric criteria mapped to each stated learning outcome.
Source bibliography
Every primary regulation and standard cited on this page, linked and dated.
Which AI agents does this course use?
The hands-on modules run against agents already deployable on the ibl.ai platform for legal.
Where does the course material come from?
Every module is grounded in primary sources — the regulation, standard, or research itself, not a summary of it. Each was resolved at authoring time.
- EDRM Model
EDRM
The lifecycle framework the course maps AI onto.
- Federal Rules of Civil Procedure Rule 26
Cornell Legal Information Institute
Proportionality and disclosure obligations analyzed in Modules 6 and 7.
- CourtListener
Free Law Project
Case law on TAR acceptance researched in Module 2.
- Model Rules of Professional Conduct
American Bar Association
Competence and supervision duties across the review process.
Delivery notes
Binding guidance for anyone preparing and delivering this course.
- Module 3 must be honest that generative privilege screening lacks TAR's judicial track record. Overstating its acceptance would put firms in a defensibility position they cannot support.
- Module 4 needs real statistics, taught properly. eDiscovery validation is frequently done badly and a course that hand-waves the sampling produces indefensible numbers.
- Use a synthetic document population with known privilege for Module 3. Measuring error rates requires ground truth that a real matter cannot provide.
- Module 6's role-play should include a sophisticated objection. Opposing counsel increasingly challenge AI methodology and participants need to have answered a hard version once.
- Have an experienced eDiscovery special master or expert review the course. Defensibility standards are set by practice and case law, not by the technology.
Why run AI training on a platform you own?
You own the course, not a licence to it
Course content, learner data, and the platform run inside your perimeter — you own all the code and the data.
Model-agnostic delivery
Run the course's AI components on any LLM — Claude, GPT, Llama, Gemini, Command — and switch anytime.
No per-seat training licences
Usage-based or self-hosted, so cost tracks actual use rather than headcount.
Deploy anywhere
Cloud, private VPC, on-premise, or fully air-gapped — including for cohorts that cannot use public AI tools.
Frequently asked questions
What does the AI in eDiscovery: TAR, Privilege Screening, and Defensibility course cover?
Technology-assisted review has judicial acceptance; generative privilege screening does not yet have the same track record. This course covers where AI changes discovery economics, the validation and sampling that make a process defensible, and how to document methodology so it survives a challenge from opposing counsel. It runs 6 hours across 8 modules across 8 modules, at advanced level, and closes with a capstone: Defensible TAR workflow with a validation report.
Who should take AI in eDiscovery: TAR, Privilege Screening, and Defensibility?
It is written for eDiscovery counsel and litigation support, Litigation partners and associates, Document review managers, In-house litigation counsel. Prerequisites: eDiscovery experience; Familiarity with the EDRM lifecycle.
Can we run this course on our own infrastructure?
Yes. ibl.ai is model-agnostic and deploy-anywhere — cloud, private VPC, on-premise, or fully air-gapped — and you own all the code and the data. Cohort data, submissions, and any material learners upload stay inside your perimeter, which matters for legal teams that cannot send work to a public AI tool.
How do we get access to AI in eDiscovery: TAR, Privilege Screening, and Defensibility?
Request access and we will set it up for your cohort — hosted by ibl.ai, or running against your own deployment. Tell us the group size and timing you need, and whether it should run inside your own perimeter.
How much does AI training for legal cost on ibl.ai?
There is no per-seat pricing — you pay for usage or self-host and pay only for the infrastructure, so a 5,000-person rollout does not cost 5,000 licences. 1.6M+ users across 400+ organizations run the platform this way, including NVIDIA, MIT, and Syracuse University.