Capstone Project Proposal
Responsible AI Integration in Legal Practice: Building Technical Competence in AI Tooling and Applications
Focus Area: Technical AI Skills Development
Intern Information
| Field | Details |
|---|---|
| Name | Isabel Budenz |
| Program | LLM International Commercial Arbitration, University of Stockholm (2025-2026) |
| Background | LLB International and European Law, University of Groningen (2022-2025) |
| Languages | German (Native), Spanish (Native), English (C2), French (B1) |
| Relevant Experience | Legal Researcher, A for Arbitration (2019-2025); Clifford Chance Antitrust Global Virtual Internship |
| Relevant Coursework | Introduction to AI and the EU AI Act; International Commercial Arbitration |
Executive Summary
Legal professionals who can bridge the gap between law and technology are increasingly valuable. 79% of law firms have adopted AI tools, yet few lawyers have formal AI training. This project focuses on building Isabel’s hands-on technical competence with AI tools while producing practical resources that help legal practitioners integrate AI responsibly.
Unlike the other project proposals that emphasize legal analysis, this project prioritizes learning by doing—working directly with AI tools, understanding their technical capabilities and limitations, and developing practical implementation guidance that meets ABA ethical standards.
This technical skills focus transforms Isabel from a legal professional who understands AI policy into one who can implement, evaluate, and govern AI systems in practice.
Problem Statement
The Legal AI Skills Gap
Adoption vs. Competence
- 79% of law firms have adopted AI tools (2024)
- Few lawyers have formal AI training
- 52% of law firm managers have shifted hiring criteria due to AI advances
- 66% of in-house legal managers seek different skills due to automation
Ethical Framework Without Practical Guidance
ABA Formal Opinion 512 (July 2024) requires lawyers to:
- Understand AI capabilities and limitations (Rule 1.1 Competence)
- Protect client information when using AI (Rule 1.6 Confidentiality)
- Keep clients informed about AI use (Rule 1.4 Communication)
- Verify AI-generated citations (Rules 3.1, 3.3 Candor)
- Establish firm-wide AI policies (Rules 5.1, 5.3 Supervision)
But: Opinion 512 provides principles, not practical implementation guidance.
State Requirements Accelerating
- New York: 2 annual CLE credits in AI competency (Q3 2025)
- Pennsylvania: Mandatory AI disclosure in court submissions
- California: Multi-jurisdictional compliance for AI cloud tools
Business Need: [Company Name] needs team members who understand AI tools practically—not just legally—to evaluate products, advise clients, and implement responsible AI practices.
Project Objectives
Primary Objectives
- Develop hands-on proficiency with 5+ legal AI tools across research, contract analysis, and drafting
- Create a comprehensive AI tool evaluation framework aligned with ABA Opinion 512 requirements
- Build a prompt engineering playbook for legal tasks with tested prompts and quality control protocols
- Develop firm-wide AI policy templates and training curriculum
Secondary Objectives (Skills Development)
- Earn AI-related certifications (Clio Legal AI Fundamentals, Coursera Prompt Engineering)
- Understand technical AI concepts (NLP, LLMs, hallucinations, bias) at practitioner level
- Build portfolio of technical artifacts demonstrating cross-disciplinary competence
Technical Learning Objectives
Conceptual Understanding
| Topic | Learning Objective |
|---|---|
| Machine Learning Basics | Understand supervised/unsupervised learning, training data, model outputs |
| Natural Language Processing | Comprehend how AI processes legal text, entity recognition, semantic analysis |
| Large Language Models | Understand transformer architecture at high level, context windows, token limits |
| AI Limitations | Deeply understand hallucinations, bias, confidentiality risks, accuracy boundaries |
| Prompt Engineering | Master techniques for effective, consistent AI outputs in legal contexts |
Practical Tool Proficiency
| Tool Category | Specific Platforms | Competency Target |
|---|---|---|
| Legal Research | Lexis+ AI, CoCounsel (Casetext) | Conduct research, verify citations, compare outputs |
| Contract Analysis | Harvey, Luminance, Ironclad | Review contracts, identify issues, generate summaries |
| Document Drafting | Claude, GPT-4, legal-specific tools | Draft legal documents with appropriate oversight |
| E-Discovery | Relativity AI, Reveal | Understand document review acceleration |
| General AI | Claude, ChatGPT, Gemini | Evaluate capabilities, understand limitations |
Certification Goals
| Certification | Provider | Timeline |
|---|---|---|
| Legal AI Fundamentals | Clio (Free) | Week 2 |
| Prompt Engineering for Law | Coursera/Vanderbilt | Week 6 |
| AI and the Law (if available) | Harvard Executive Ed | Post-project |
Research Foundation
Current Legal AI Landscape
Market Impact (2024-2025)
- 9% increase in legal research AI usage
- 17% increase in contract analysis (in-house)
- 34% jump in case law summarization
- 65% reduction in review time reported
- 85% decrease in human error
- 40% cost reduction
Leading Tools
| Category | Tool | Key Features |
|---|---|---|
| Research | Lexis+ AI | Natural language queries, citation verification |
| Research | CoCounsel | GPT-4 powered, deposition prep, timeline creation |
| Contracts | Harvey | Generative AI for law firms, M&A due diligence |
| Contracts | Luminance | ML document review, anomaly detection |
| Contracts | Ironclad | CLM with AI assistant, redline generation |
| E-Discovery | Relativity | AI-powered review, privilege detection |
| General | Claude | Long context, nuanced analysis, safety focus |
Ethical Requirements
ABA Opinion 512 Core Requirements
- Competence: Understand capabilities AND limitations
- Confidentiality: Assess data handling, opt out of training where possible
- Communication: Inform clients of AI use in their matters
- Candor: Independently verify all AI outputs
- Supervision: Establish policies, train staff, monitor use
Key Risk Areas
- Hallucinations (fabricated citations, false facts)
- Confidentiality breaches (data used for training)
- Bias in outputs (training data limitations)
- Over-reliance (failure to verify)
- Unauthorized practice (AI providing legal advice)
Scope
In Scope
| Area | Details |
|---|---|
| Tools | 5+ legal AI platforms across research, contracts, drafting |
| Tasks | Legal research, contract review, document drafting, due diligence |
| Frameworks | ABA Opinion 512, state-specific requirements (NY, CA, PA) |
| Outputs | Evaluation framework, prompt playbook, policy templates, training |
Out of Scope
- AI tool development or coding
- Deep technical ML/AI research
- Vendor negotiations or procurement
- Client-facing AI implementation
Deliverables
| # | Deliverable | Description | Format | Due |
|---|---|---|---|---|
| 1 | AI Tool Proficiency Log | Documented hands-on experience with 5+ tools, including outputs and assessments | Portfolio Document (30+ pages) | Ongoing → Week 10 |
| 2 | AI Tool Evaluation Framework | Criteria and methodology for assessing legal AI tools against ethical requirements | Framework (20 pages) + Scorecard Template | Week 5 |
| 3 | Prompt Engineering Playbook | Tested prompts for common legal tasks with quality control protocols | Playbook (40+ pages) + Prompt Library | Week 8 |
| 4 | ABA Opinion 512 Compliance Checklist | Practical checklist mapping ethical requirements to operational practices | Checklist + Implementation Guide | Week 9 |
| 5 | Firm-Wide AI Policy Templates | Model policies for AI use, data handling, disclosure, supervision | Policy Templates + Adoption Guide | Week 11 |
| 6 | Training Curriculum & Materials | Complete training program for legal professionals on responsible AI use | Curriculum + Slides + Exercises | Week 12 |
Certification Deliverables
| Certification | Evidence | Timeline |
|---|---|---|
| Clio Legal AI Fundamentals | Certificate | Week 2 |
| Prompt Engineering for Law | Certificate | Week 6 |
Methodology
Phase 1: Foundation Building (Weeks 1-3)
Week 1: Conceptual Learning
- Complete Clio Legal AI Fundamentals certification
- Study ML/NLP basics through curated resources
- Understand LLM architecture at practitioner level
- Document learning in proficiency log
Week 2: Ethics Deep Dive
- Analyze ABA Opinion 512 comprehensively
- Review state-specific AI requirements
- Study documented AI failures in legal contexts
- Begin drafting evaluation framework criteria
Week 3: Initial Tool Exploration
- Obtain access to target AI tools
- Conduct initial exploration of each platform
- Document capabilities, interfaces, limitations
- Begin systematic testing protocol
Phase 2: Hands-On Tool Mastery (Weeks 4-6)
Week 4: Legal Research Tools
- Deep dive into Lexis+ AI and CoCounsel
- Test with real-world research scenarios
- Compare outputs, verify accuracy
- Document hallucination rates, citation accuracy
- Update proficiency log with detailed findings
Week 5: Contract Analysis Tools
- Explore Harvey, Luminance, or Ironclad
- Test contract review capabilities
- Assess issue identification accuracy
- Evaluate redline and summary features
- Complete AI Tool Evaluation Framework
Week 6: Prompt Engineering Mastery
- Complete Coursera Prompt Engineering certification
- Develop and test prompts for common legal tasks:
- Legal research queries
- Contract review instructions
- Document drafting prompts
- Due diligence checklists
- Document effective techniques and failures
Phase 3: Framework Development (Weeks 7-9)
Week 7-8: Prompt Playbook Development
- Compile tested prompts into organized playbook
- Develop quality control protocols for each task type
- Create prompt templates with variables
- Document edge cases and failure modes
- Complete Prompt Engineering Playbook
Week 9: Compliance Implementation
- Map ABA Opinion 512 to practical operations
- Develop checklist for each ethical requirement
- Create workflow integration guidance
- Complete ABA Opinion 512 Compliance Checklist
Phase 4: Policy & Training Development (Weeks 10-12)
Week 10: Policy Template Creation
- Draft firm-wide AI use policy
- Develop data handling and confidentiality protocols
- Create disclosure templates (client, court)
- Build supervision and monitoring framework
- Finalize AI Tool Proficiency Log
Week 11: Policy Refinement
- Review policies with mentor and legal team
- Incorporate feedback
- Develop adoption roadmap
- Complete Firm-Wide AI Policy Templates
Week 12: Training Program Development
- Design training curriculum structure
- Create presentation materials
- Develop hands-on exercises
- Pilot training session
- Complete Training Curriculum & Materials
Timeline
Week 1 ████░░░░░░░░░░░░░░░░░░░░ Foundation: Clio cert + conceptual learning
Week 2 ████░░░░░░░░░░░░░░░░░░░░ Ethics deep dive + evaluation criteria
Week 3 ████░░░░░░░░░░░░░░░░░░░░ Initial tool exploration
Week 4 ░░░░████░░░░░░░░░░░░░░░░ Legal research tools mastery
Week 5 ░░░░████░░░░░░░░░░░░░░░░ Contract tools + Evaluation Framework
Week 6 ░░░░░░░░████░░░░░░░░░░░░ Prompt engineering cert + testing
Week 7-8 ░░░░░░░░░░░░████████░░░░ Prompt Playbook development
Week 9 ░░░░░░░░░░░░░░░░████░░░░ Compliance Checklist
Week 10-11 ░░░░░░░░░░░░░░░░░░░░████ Policy Templates + Proficiency Log
Week 12 ░░░░░░░░░░░░░░░░░░░░░░██ Training Curriculum + Delivery
Skills Development Tracking
Technical Skills Matrix
| Skill | Starting Level | Target Level | Assessment Method |
|---|---|---|---|
| ML/NLP Concepts | Novice | Practitioner | Quiz + explanation exercise |
| Prompt Engineering | Novice | Proficient | Playbook quality + cert |
| Tool Proficiency (Research) | Novice | Proficient | Task completion + accuracy |
| Tool Proficiency (Contracts) | Novice | Intermediate | Task completion + evaluation |
| AI Risk Assessment | Intermediate | Advanced | Framework quality |
| Training Delivery | Intermediate | Proficient | Pilot session feedback |
Weekly Skill Check-ins
Each week includes:
- Learning log: What was learned, what remains unclear
- Tool hours: Time spent with each AI tool
- Prompt experiments: Prompts tested, results documented
- Failure documentation: What didn’t work and why
Resources Required
Tool Access
| Tool | Access Type | Priority |
|---|---|---|
| Claude Pro | Subscription | Week 1 |
| Lexis+ AI | Firm subscription | Week 3 |
| CoCounsel/Casetext | Trial or subscription | Week 3 |
| Harvey | Demo access | Week 5 |
| Luminance | Trial | Week 5 |
Learning Resources
| Resource | Provider | Cost |
|---|---|---|
| Legal AI Fundamentals | Clio | Free |
| Prompt Engineering for Law | Coursera | ~$50 |
| AI and the Law readings | Various | Provided |
| ABA Opinion 512 + commentary | ABA | Free |
Subject Matter Expert Support
| Role | Purpose | Time |
|---|---|---|
| Primary Mentor | Weekly guidance | 2 hrs/week |
| Technology Counsel | AI tool expertise | 4 hrs total |
| Training Specialist | Curriculum development | 3 hrs total |
| IT/Security | Data handling review | 2 hrs total |
Budget
| Item | Estimated Cost |
|---|---|
| Tool subscriptions/trials | $500 |
| Certification courses | $100 |
| Learning materials | $100 |
| Total | $700 |
Success Criteria
Skills Acquisition
- Clio Legal AI Fundamentals certification earned
- Prompt Engineering certification completed
- 50+ hours logged with AI tools
- Proficiency demonstrated in 5+ platforms
- Can explain ML/NLP concepts accurately
Deliverable Quality
- All 6 deliverables completed on schedule
- Prompt playbook contains 30+ tested prompts
- Evaluation framework validated by technology counsel
- Policy templates approved by compliance team
- Training pilot receives >4/5 feedback
Business Impact
- Framework adopted for tool evaluation
- Policies implemented firm-wide
- Training delivered to 20+ professionals
- At least 2 tool recommendations accepted
Risks and Mitigation
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| Tool access delays | Medium | High | Identify alternatives; prioritize widely available tools |
| Learning curve steeper than expected | Medium | Medium | Build buffer time; focus on breadth over depth |
| Rapid tool evolution during project | Medium | Low | Focus on principles; note tool-specific vs. generalizable learnings |
| Certification scheduling conflicts | Low | Low | Complete early; identify alternatives |
| Confidentiality concerns with testing | Medium | High | Use synthetic/public data only; follow firm protocols |
Career Positioning Value
This project transforms Isabel into a legally-trained AI practitioner:
Differentiators
| Traditional Legal Professional | Isabel After This Project |
|---|---|
| Understands AI regulation | Can evaluate and implement AI tools |
| Reads about AI capabilities | Has hands-on proficiency with platforms |
| Knows ethical requirements exist | Can operationalize ABA Opinion 512 |
| Aware of prompt engineering | Has tested prompt library for legal tasks |
| Understands training needs | Can deliver AI training programs |
Career Paths Enabled
- Legal Engineer: Bridge law and technology teams
- AI Implementation Lead: Guide firm AI adoption
- Legal Tech Product Counsel: Advise AI tool development
- In-House AI Governance: Oversee responsible AI use
- Consultant: Help firms implement legal AI
Portfolio Assets
- Certifications: Demonstrable AI competence
- Prompt Playbook: Practical, tested resource
- Evaluation Framework: Methodology for tool assessment
- Policy Templates: Ready-to-implement governance
- Training Materials: Delivery capability demonstrated
Alignment with Industry Trends
| Trend | Project Relevance |
|---|---|
| 79% law firm AI adoption | Proficiency makes Isabel immediately valuable |
| ABA Opinion 512 compliance pressure | Checklist and policies address urgent need |
| NY AI CLE requirement (2025) | Training curriculum directly applicable |
| Prompt engineering as “21st-century legal skill” | Playbook demonstrates mastery |
| Legal engineer role emergence | Technical + legal competence combination |
Stakeholders
| Stakeholder | Role | Engagement |
|---|---|---|
| Primary Mentor | Day-to-day guidance | Weekly 1:1 |
| Technology Counsel | Tool expertise, evaluation validation | Bi-weekly |
| Training/Professional Development | Curriculum review | Weeks 10-12 |
| IT/Security | Data handling, tool vetting | Ad hoc |
| Legal Teams | Policy feedback, training participants | Weeks 9-12 |
Approval
Intern Acknowledgment
I have reviewed this proposal and commit to delivering the outlined project within the specified timeline and quality standards. I understand this project emphasizes hands-on technical skill development alongside traditional legal analysis.
Intern Signature: _________ Date: _____
Isabel Budenz
Mentor Approval
Mentor Signature: _________ Date: _____
Executive Sponsor Approval
Sponsor Signature: _________ Date: _____
| *Proposal Version 1.0 | Focus: Technical AI Skills Development | January 2026* |