事件
Harrison & Cole LLP had been talking about AI for two years. Founded in 1987 as a mid-size firm specializing in corporate law, intellectual property, and commercial litigation, the firm had a reputation for technical excellence and measured decision-making. But 'measured' was starting to feel like 'slow.' Three of the firm's top ten clients had asked in the past quarter whether the firm was 'using AI yet.' Two lateral partner candidates had declined offers, citing the firm's technology infrastructure as 'behind the curve.'
Managing partner Victoria Harrison finally drew the line. She formed an AI Selection Committee with a mandate, a budget, and a deadline: evaluate available AI solutions, select one platform, and present a recommendation to the full partnership within 60 days. The budget was $200,000 for the first year, including license fees, implementation, and training. 'I don't want a report,' she told the committee. 'I want a decision.'
The committee issued an RFP and received responses from eleven vendors. After initial screening, three finalists emerged — each offering a fundamentally different value proposition, and each backed by a compelling pitch. The committee now had 30 days to evaluate, test, and choose. What followed was a case study in how organizational priorities collide when technology meets legal practice.
主要タイムライン
Day 1-15 — RFP and Initial Screening
The committee issues an RFP to 20 vendors. Eleven respond. The committee screens for baseline requirements: SOC 2 certification, data residency in the United States, GDPR compliance capability, and integration with the firm's existing iManage DMS. Three vendors pass all threshold criteria and are invited to present.
Day 16-25 — Vendor Presentations and Demos
Each vendor conducts a 90-minute presentation and live demo for the committee. Vendor A (DocuLex AI) emphasizes breadth — a full-suite platform covering research, drafting, review, and matter management. Vendor B (PrecisionLegal) focuses on depth — a specialized litigation support tool with advanced e-discovery and predictive analytics. Vendor C (CounselMind) offers an AI-native platform purpose-built for corporate transactional work with strong contract intelligence capabilities.
Day 26-45 — Hands-On Evaluation
Each vendor provides a 14-day trial environment. Committee members and selected attorneys test the tools using redacted versions of actual firm documents. Results are mixed: DocuLex AI handles breadth well but lacks depth in any single area. PrecisionLegal excels at litigation tasks but is irrelevant for the corporate and IP practices. CounselMind produces impressive contract analysis but has a steeper learning curve and limited litigation capability.
Day 46-60 — Decision Phase
The committee must reconcile conflicting evaluation results, budget constraints (CounselMind is $60,000 over budget at full implementation), a security concern flagged by the DPO regarding DocuLex AI's European data processing sub-processors, and internal politics — the litigation practice group leader is lobbying hard for PrecisionLegal while the corporate group wants CounselMind.
なぜこれが重要か
AI vendor selection is not a technology decision — it is an organizational strategy decision that will shape the firm's capabilities, competitive position, and risk profile for years to come. There is no objectively 'right' answer among the three vendors. The right answer depends on the firm's strategic priorities, risk tolerance, budget discipline, and ability to manage change. Learning to navigate this ambiguity — with incomplete information, competing stakeholder interests, and real financial constraints — is one of the most important skills a legal professional can develop in the AI era.
コンテキスト分析
The technical, financial, organizational, and regulatory dimensions that shape the selection process.
Vendor Comparison: DocuLex AI (Full Suite)
- Covers research, drafting, document review, and matter management in a single platform — attractive for firm-wide adoption
- Pricing: $180,000/year enterprise license (within budget)
- Strengths: Breadth of coverage, strong iManage integration, 24/7 support, established market presence with 200+ law firm clients
- Weaknesses: 'Jack of all trades' — none of its individual modules match the depth of specialized competitors. European data processing sub-processors raise GDPR questions for the firm's EU-based clients
Vendor Comparison: PrecisionLegal (Litigation Specialist)
- Deep litigation support: e-discovery, predictive coding, deposition analysis, outcome prediction, trial preparation
- Pricing: $140,000/year (well within budget, leaving room for future expansion)
- Strengths: Best-in-class litigation analytics, proven accuracy in e-discovery (validated by independent TREC study), strong track record in AmLaw 200 firms
- Weaknesses: Zero capability for corporate, IP, or transactional work — serving only 40% of the firm's practice. No contract review, no drafting assistance
Vendor Comparison: CounselMind (Corporate/Transactional)
- Purpose-built for corporate transactions: contract intelligence, M&A due diligence, regulatory compliance monitoring, clause library
- Pricing: $260,000/year full implementation ($60,000 over budget); $195,000/year for a reduced-scope deployment
- Strengths: Superior contract analysis accuracy (92% in independent testing), AI-native architecture (not a bolt-on), strong data privacy controls with on-premise option
- Weaknesses: Steeper learning curve (estimated 20 hours training per user vs. 8 hours for DocuLex), limited litigation capability, smaller company with fewer support resources
Organizational Context
- Harrison & Cole's revenue split: 40% litigation, 35% corporate/transactional, 25% IP — no single practice group dominates
- The firm's strategic plan prioritizes growth in corporate advisory services over the next 3 years
- Attorney technology adoption rates vary: corporate attorneys tend to be early adopters, litigation attorneys are mixed, IP attorneys are generally skeptical
- The firm's last major technology investment (a practice management system) took 18 months to reach full adoption and came in 40% over the initial budget
ステークホルダーと役割
Each participant assumes one role on the AI Selection Committee and advocates for their perspective throughout the evaluation process.
Diane Reeves — Chief Technology Officer
プロフィール
Joined the firm two years ago from a legal technology vendor. Deep technical expertise but still building credibility with the partnership. Responsible for the RFP process and technical evaluation. Focused on architecture, integration, scalability, and long-term technology strategy.
目的
- Recommend the solution with the strongest technical architecture and best long-term scalability
- Ensure seamless integration with the firm's existing iManage DMS and billing systems
- Establish the CTO role as the firm's authoritative voice on technology decisions
制約
Diane privately favors CounselMind's AI-native architecture but knows the budget overrun will be a hard sell. She is also aware that her previous employer was a competitor to DocuLex AI, which could create a perception of bias if she argues against it too forcefully.
Marcus Webb — Litigation Practice Group Leader
プロフィール
Twenty-year veteran litigator and the firm's highest-revenue partner. His practice group generates 40% of firm revenue. Politically influential and accustomed to getting what he wants. Believes the firm's AI investment should prioritize the practice group that generates the most revenue.
目的
- Secure a tool that directly enhances his practice group's litigation capabilities, particularly e-discovery and predictive analytics
- Prevent the firm from investing $200,000 in a tool that provides no benefit to 40% of its revenue
- Maintain the litigation group's influence over firm-wide technology decisions
制約
Marcus knows that PrecisionLegal serves only his practice group and that advocating for it looks self-serving. He is also aware that two of the three clients who asked about AI capabilities are corporate clients, not litigation clients — a fact he would prefer not to highlight.
Priya Sharma — Finance Director
プロフィール
The firm's finance director for eight years. Conservative with budget allocation and deeply skeptical of technology vendor projections. Her job is to protect the firm's financial health and ensure that investments deliver measurable returns. She evaluates everything through an ROI lens.
目的
- Keep the total first-year cost within the $200,000 budget — no exceptions, no 'phase 2' budget expansions
- Require each vendor to provide verifiable ROI projections with specific, measurable success criteria
- Build a vendor contract that includes performance guarantees and termination provisions
制約
Priya's budget analysis shows that CounselMind's reduced-scope deployment ($195,000) is technically within budget but leaves almost no contingency for implementation costs. She estimates true first-year costs at 130-150% of the license fee based on the firm's previous technology adoption experience.
Oliver Grant — Data Protection Officer
プロフィール
Recently appointed DPO responsible for the firm's data protection compliance. Background in privacy law with certifications in GDPR and CCPA. Evaluates every technology decision through a data protection lens. His approval is required before any tool processing client data can be deployed.
目的
- Ensure that the selected vendor's data handling practices comply with all applicable data protection regulations and the firm's client engagement letter requirements
- Conduct a Data Protection Impact Assessment (DPIA) for the selected tool before deployment
- Establish data protection requirements as non-negotiable threshold criteria, not factors to be weighed against other considerations
制約
Oliver has identified a specific concern with DocuLex AI: the vendor uses European sub-processors for certain AI inference tasks, which may create data transfer issues under the firm's engagement letters with three clients that require all data processing to remain within the United States. He is still assessing whether CounselMind's on-premise option fully resolves data residency concerns.
学習アクティビティ
Six task categories based on the Smoother methodology, designed to build progressively deeper understanding of the AI tool selection process.
- Create a detailed comparison matrix of the three vendors across technical capability, pricing, security, integration, support, and strategic fit dimensions.
- Research each vendor's market position: How long have they been operating? How many law firm clients do they serve? Have they received independent validation of their accuracy claims?
- List all the stakeholders who are affected by this decision — not just the committee members, but attorneys, paralegals, clients, and the firm's competitors. Map their interests.
- Identify the specific data protection regulations and client engagement letter requirements that constrain the selection. Which vendors comply, which do not, and which are uncertain?
- Explain why each committee member favors a different vendor. What professional interests, risk tolerances, and strategic visions drive each preference?
- Restate the selection problem from three different angles: as a technology decision, as a financial decision, and as a strategic positioning decision. How does the framing change the analysis?
- Interpret the trial results: what do they actually tell us about each tool's real-world performance? What are the limitations of a 14-day evaluation with redacted documents?
- Analyze the relationship between the firm's revenue distribution (40% litigation, 35% corporate, 25% IP) and the vendor selection. Should revenue share determine technology investment?
- Challenge the assumption that the firm must choose a single vendor. Is a multi-vendor strategy viable? What are its advantages and risks?
- Evaluate whether the $200,000 budget is adequate for meaningful AI adoption at a 60-attorney firm. What would a realistic budget look like, and what should be cut or deferred if the budget cannot be increased?
- Assess the risk of selecting a specialized tool (PrecisionLegal or CounselMind) that serves only part of the firm versus a generalist tool (DocuLex AI) that serves all practice groups adequately but none excellently.
- Question CounselMind's 92% accuracy claim. What does '92% accuracy' mean in the context of contract analysis? Accuracy at what task? Measured how? Against what baseline?
- Draft the committee's recommendation memo to the full partnership. Include: the recommended vendor, the rationale, the risks, the budget analysis, the implementation timeline, and the success criteria for the first year.
- Design a phased implementation plan for the selected vendor that manages risk, builds internal capability, and includes clear go/no-go decision points.
- Negotiate a vendor contract: draft the key terms you would require, including performance guarantees, data handling provisions, termination clauses, and pricing protections for years 2 and 3.
- Create a communication plan for the firm: how will the selection decision be announced, how will training be organized, and how will resistance from practice groups that did not get their preferred tool be managed?
- Peer-review another group's recommendation memo. Does the rationale hold up under scrutiny? Are the risks adequately addressed? Would you vote for this recommendation as a partner?
- Evaluate the decision-making process itself: Did the committee structure produce a good decision? What biases, power dynamics, or information gaps affected the outcome?
- Compare your implementation plan with others. Which approach best manages the risk of a technology investment that fails to deliver expected returns?
- Assess whether the firm's AI selection process could serve as a template for future technology decisions. What would you change for the next evaluation?
- Reflect on your own decision-making process during this case study. Did you favor a vendor early and then look for evidence to support that preference? How did you manage confirmation bias?
- What assumptions about AI tools did you bring into this exercise? Which were validated and which were challenged?
- Consider how organizational politics affected the committee's process. In your experience, how do firms actually make technology decisions — and how should they?
- What is the most important lesson from this case study for your own practice or organization?
実践に活かす
Using the evaluation framework developed in this case study, assess one AI tool that is relevant to your current practice. Issue a mock RFP to three vendors, compare their responses using weighted criteria, and document your recommendation with a rationale that addresses technical capability, financial impact, security compliance, and strategic fit.
参考文献・出典
調達と評価
- ILTA (International Legal Technology Association), "AI Procurement Framework for Law Firms" — structured evaluation methodology
- ACC (Association of Corporate Counsel), "Model AI Procurement Clauses" — contract provisions for AI vendor agreements
- Gartner, "Critical Capabilities for AI in Legal" — methodology for evaluating AI tools against specific use cases
データ保護とコンプライアンス
- ABA Formal Opinion 477R — Securing Communication of Protected Client Information in the context of technology adoption
- GDPR Articles 28 and 44-49 — Data processor requirements and international data transfer provisions
- NIST AI Risk Management Framework — structured approach to identifying and managing AI-related risks
AIツール選定をマスターする準備はできていますか？
This case study is part of Module 2 of the Lawra Learning Program. Request a facilitated session with role assignments, vendor simulation materials, and expert guidance tailored to your firm's specific needs.
コメント
コメントを読み込み中...