About this Claude for Legal Teams course
Legal teams work with large volumes of contracts, correspondence, research and matter information, often where small errors can have significant consequences.
This practical one-day course explores how Claude can assist with everyday legal work while retaining the professional judgement, verification and supervision that legal practice requires.
Using fictional legal documents throughout, you will use Claude to extract and compare contract terms, review clauses against a playbook, redraft clauses and correspondence, build chronologies and support legal research. You will learn how to make outputs easier to verify by requiring source and clause references, distinguishing source material from inference, and checking authorities against authoritative legal sources.
The course also examines confidentiality, privilege, client information and the risks associated with untrusted documents. You will consider how repeatable workflows can be governed using appropriate human review, records and controls.
The course teaches practical use of Claude within legal work. It does not provide legal advice, and Claude is not treated as an authoritative source of law.
What you will need
A laptop with a current web browser
Access to Microsoft Word
An appropriate Claude account supporting the features used during the course
Permission from their organisation to use that account for the course exercises
Access to an authoritative legal research source where available
The course uses fictional contracts, matter information and research materials. Real client, matter, privileged or confidential information should not be used during the exercises.
Some exercises may demonstrate integrations or Claude features that depend on the delegate's account, subscription or organisation settings. Where a feature is unavailable, the instructor can demonstrate the relevant workflow.
-
By the end of this course, attendees will be able to:
- Identify legal tasks where Claude can usefully assist and recognise activities that require professional judgement and human control
- Write structured prompts that specify the jurisdiction, client position, required output and permitted source material
- Extract information from agreements with clause references that can be checked against the original
- Use Claude to assist with contract review against an agreed playbook
- Compare document versions and identify significant changes for professional review
- Use Claude to support clause, correspondence and client-communication drafting
- Use AI to support legal research while independently verifying cases, legislation and quotations against authoritative sources
- Build chronologies and issue lists with traceable source references
- Apply appropriate precautions when working with confidential, privileged or client information
- Recognise risks presented by hidden or malicious instructions in documents
- Create repeatable Claude workflows while retaining appropriate supervision, records and review
-
This course is intended for people in law firms, in-house legal teams and legal support functions who want to use Claude more effectively while retaining appropriate professional oversight.
It will be particularly useful for:
Solicitors and in-house counsel working with commercial agreements
Paralegals, legal executives and trainees
Legal operations and practice-support teams
Contract managers and commercial teams responsible for reviewing agreements
Partners, general counsel and legal-team leaders responsible for the appropriate use of AI
No previous experience of Claude is required.
-
Delegates should:
Be familiar with commercial contracts and the way their team reviews them
Be comfortable working with Microsoft Word
Understand their organisation's requirements for handling confidential and client information
Previous experience of an AI assistant is helpful but not required.
-
This Claude for Legal Teams course is available for private / custom delivery for your team - as an in-house face-to-face workshop at your location of choice, or as online instructor-led training via MS Teams (or your own preferred platform).
Get in touch to find out how we can deliver tailored training which focuses on your project requirements and learning goals.
-
Claude in legal practice
Appropriate uses of generative AI within legal work
Common failure modes, including incorrect quotations, missed cross-references and invented authorities
Keeping professional responsibility with the practitioner
Deciding whether Claude should draft or assist, or whether a task should remain entirely with a person
Practical exercise: Classify common legal activities according to the role Claude and the supervising professional should play.
Confidentiality and controlled use
Client and confidential information
Legal professional privilege
Approved organisational accounts and services
Understanding how information is handled by the chosen service
Risks associated with documents received from third parties
Hidden instructions in comments, tracked changes, headers and other document content
Following your organisation's policies and controls
Practical exercise: Apply a fictional firm's AI-use policy to different types of matter information and determine the permitted treatment of each.
Prompting for legal work
Establishing the goal and intended audience
Specifying jurisdiction and governing law
Identifying which party you act for
Constraining Claude to supplied source material
Requesting clause references and supporting text
Separating fact, inference and uncertainty
Asking Claude to identify information that is absent rather than inventing an answer
Practical exercise: Turn a broad request to review an NDA into a structured legal-work prompt and compare the results.
Contract extraction and verification
Extracting key commercial terms
Working with defined terms, schedules, tables and cross-references
Requiring clause references and supporting wording
Identifying information that is absent or uncertain
Checking extracted information against the source document
Practical exercise: Extract key terms from a fictional services agreement and verify a selection against the original contract.
Reviewing contracts against a playbook
Using an agreed playbook rather than relying on general model knowledge
Comparing clauses with preferred, fallback and unacceptable positions
Identifying deviations for professional review
Comparing agreement versions and counterparty changes
Assessing which changes warrant closer attention
Practical exercise: Review a fictional supplier agreement against a supplied playbook and compare two versions.
Drafting and redrafting
Redrafting contract clauses
Maintaining defined terms and consistency
Supporting correspondence and short advice-note drafting
Producing plain-English explanations
Reviewing every proposed change before acceptance
Using supported Microsoft Word integrations where available
Practical exercise: Redraft a fictional contractual provision and review the proposed changes before accepting or rejecting them.
Research support and authority checking
Using Claude to help frame research questions and identify search terms
Why generative AI should not be treated as a source of law
Checking whether an authority exists
Confirming citations and quotations
Checking jurisdiction and currency
Verifying propositions against authoritative legal sources
Practical exercise: Check the authorities contained in a fictional research note and identify unreliable or unsupported material.
Chronologies, issue lists and communications
Building chronologies from multiple documents
Retaining source references for individual entries
Identifying inconsistencies and unresolved points
Drafting plain-English client updates
Separating facts, supplied legal advice and suggested next steps
Checking commitments, deadlines and instructions before sending
Practical exercise: Build and verify a chronology from a fictional matter bundle before drafting a client update.
Creating repeatable team workflows
Using Projects to provide reference material and standing instructions
Capturing recurring tasks using reusable Claude features
Working with team playbooks, precedents and house style
Ownership and version control of the authoritative source material
Retesting workflows when instructions, source material or AI functionality changes
Practical exercise: Design a reusable workflow for reviewing fictional NDAs against an agreed playbook.
Supervision, evidence and risk management
Applying existing professional responsibilities to AI-assisted work
Establishing permitted and prohibited uses
Deciding what must be independently checked
Identifying appropriate reviewers
Recording sources, checks and approvals
Reviewing AI-assisted workflows as tools and organisational requirements change
Practical exercise: Create a simple control sheet covering permitted use, required checks, reviewer, evidence and review date.
Course outcome
By the end of the day, you will have worked through a fictional legal matter using Claude to support:
Contract extraction and review
Comparison against a playbook
Redrafting
Authority checking
Chronology creation
Client communication
You will have checked the resulting work against its source material and developed a control sheet for a repeatable AI-assisted legal workflow.
-
Does this course teach Claude to provide legal advice?
No. The course teaches legal professionals how Claude can assist with activities such as document analysis, drafting, contract comparison and research preparation. Legal judgement, advice and responsibility remain with appropriately qualified professionals.
Can Claude be relied upon for legal research?
Claude can assist with research planning, explaining concepts and identifying possible search terms, but it should not be treated as an authoritative source of law. The course teaches a verification routine in which cases, legislation, citations and quotations are checked against authoritative legal sources before they are relied upon.
Why does the course place so much emphasis on checking AI-generated authorities?
Generative AI can produce convincing but nonexistent cases, citations and quotations. The High Court considered this problem directly in Ayinde v London Borough of Haringey and Al-Haroun v Qatar National Bank, where the Divisional Court examined false information arising from actual or suspected use of generative AI and stressed lawyers' professional responsibilities and duties to the court (see judiciary.uk).
Can we use real client documents during the course?
No. The exercises use fictional contracts, correspondence and matter information. Delegates should not upload real client, confidential, privileged or commercially sensitive material during the course. Your organisation's own information-handling and AI policies remain authoritative.
Does the course cover confidentiality and privilege?
Yes. You will consider confidential and privileged information, approved organisational accounts, data handling and the risks associated with documents received from third parties. The course teaches delegates to apply their organisation's policies rather than assume that a particular category of information is safe to upload.
Does the course cover contract drafting?
Yes. You will use Claude to assist with redrafting clauses, correspondence and other short legal material. The emphasis is on producing a draft for professional review rather than treating AI-generated wording as a finished legal product.
Can Claude review contracts against our own legal playbook?
The course demonstrates the approach using a fictional playbook containing preferred, fallback and unacceptable positions. Claude can assist in comparing contract wording with supplied criteria, but the playbook remains the source and a suitably qualified person decides what position to accept.
Is this suitable for in-house legal teams as well as law firms?
Yes. The examples focus heavily on contracts, correspondence, review workflows and controlled use, making the course relevant to in-house counsel, legal operations and commercial-contract teams as well as private practice.
Do I need an existing Claude subscription?
Delegates need access to an appropriate Claude account for the practical exercises. Certain integrations and features demonstrated during the course may depend on the account, subscription or organisation configuration available at the time of delivery.
What safeguards does the course recommend when legal teams introduce AI?
Rather than prescribing one universal control model, the course helps delegates consider permitted uses, prohibited uses, required checks, appropriate reviewers, evidence to retain and review dates. These should sit within the firm's or organisation's own professional, security, data-protection and risk-management arrangements.
-
The Law Society: Using artificial intelligence in legal practice
This October 2026 practice note is probably the single best companion resource. It addresses professional duties, AI governance, confidentiality, privilege, data protection, insurance, cyber security, training and supervision, including considerations for both law firms and in-house teams. The Law Society makes clear that it represents good-practice guidance rather than legal advice.
Solicitors Regulation Authority: Misuse of AI warning notice
Current SRA guidance specifically identifies inaccurate AI-generated information and client confidentiality as areas of concern, while stating that firms can use AI provided they continue to meet the regulator's standards.
Bar Standards Board: Guidance on the use of Artificial Intelligence and Other Technologies
Particularly useful for barristers and mixed legal audiences. The current guidance covers competence, adoption and governance of technology, responsibilities when using AI for client work, confidentiality and the implications of other parties using AI.
Solicitors Regulation Authority: Compliance tips for AI and technology
A practical complement to the warning notice, particularly for legal-team leaders. It covers assessing tools before adoption, governance, senior oversight, regulatory considerations and keeping clients' interests central to technology use.
Trusted by
