**90 days

90 days. That's how long it takes to go from zero AI tools to a measurable, working deployment in a legal department — if you follow a structured playbook. Most departments take 12-18 months because they committee the process to death.

The legal ops teams that deploy AI fastest share three traits: they pick one use case instead of five, they define success metrics before buying anything, and they treat change management as a first-class workstream — not an afterthought. Here's the 90-day playbook that works, broken into vendor selection, implementation, training, and measurement.

Days 1-15: Define the Problem Before You Shop for Solutions

The first two weeks aren't about technology. They're about documenting your pain points with enough specificity to evaluate vendors objectively. Step 1 — Audit your current state. How many legal requests come in per month? What's the average response time? What percentage of attorney time goes to routine versus strategic work? Where are the bottlenecks? Get numbers, not impressions. Survey attorneys and key business unit stakeholders. Review matter data from the past 12 months. Step 2 — Pick ONE primary use case. The most successful implementations start with a single, high-volume workflow. Contract review, legal intake automation, and outside counsel invoice review are the three safest starting points because they're high-volume, easily measured, and low-risk. Don't try to solve three problems simultaneously. Step 3 — Define success criteria. Write down three specific metrics with target numbers. Example: 'Reduce average NDA review time from 3 days to 1 day. Process 50+ contracts through the platform per month. Achieve 90%+ first-pass approval rate.' These criteria become your vendor evaluation scorecard and your post-implementation measurement framework.

Days 16-35: Vendor Selection Without the Paralysis

Most legal departments evaluate too many vendors for too long. Limit your evaluation to 3 vendors maximum. Here's how. Week 3 — Create a shortlist. Based on your primary use case, identify the three leading platforms. For intake automation: Checkbox, Tonkean, and one alternative. For contract management: Ironclad, Juro, or Agiloft depending on your size and complexity. For spend management: Brightflag, Apperio, or SimpleLegal. Week 4 — Structured demos. Don't let vendors run their standard demo. Prepare a scenario based on your actual workflow and ask each vendor to demonstrate how their tool handles it. Bring the attorneys who'll use the tool to the demos — not just legal ops and IT. Week 5 — Security and procurement. Run your security questionnaire in parallel with demos, not sequentially. Most legal AI vendors have SOC 2 Type II certification and standard security documentation ready to go. The security review kills timelines when it starts after vendor selection instead of running alongside it. Decision criteria that matter: ease of configuration (can legal ops manage it without IT?), integration with your existing tools (Slack, Teams, Salesforce, matter management system), vendor's legal-specific expertise (legal-first vendors outperform general-purpose platforms), and total cost of ownership including implementation support.

Days 36-60: Implementation That Doesn't Stall

Implementation is where most deployments die. They stall because nobody owns the project, the scope creeps, or IT involvement creates a dependency chain. Assign a named owner. One person — ideally the legal ops lead — owns the implementation timeline, vendor relationship, and weekly status updates. Not a committee. Not a shared responsibility. One person who's accountable for go-live on day 60. Configure in phases. Phase 1 (days 36-45): set up the platform, configure your primary use case workflow, and migrate relevant templates or data. Phase 2 (days 46-55): integrate with existing tools (email, Slack/Teams, matter management system). Phase 3 (days 56-60): UAT testing with 3-5 pilot users from the attorney team. Avoid scope creep. You'll discover 10 additional workflows you want to automate during implementation. Write them down. Don't configure them. They're phase 2, after your primary use case is live and measured. The biggest implementation risk isn't technical — it's organizational. When a senior partner says 'can it also do X?' the answer is 'yes, in phase 2.' Every feature added before go-live delays go-live.

Days 61-75: Training and Change Management

The tool is live. Now you need people to use it. Training isn't a one-time event. It's a three-part program. Session 1 — Launch training (day 61-63). 30-minute live demo for all attorneys. Show the tool solving a real workflow they deal with weekly. Record it for asynchronous viewing. Session 2 — Practice group workshops (days 64-70). Hands-on sessions with each practice group using their specific contract types, matter categories, or request workflows. This is where attorneys see how the tool applies to their work, not a generic demo. Session 3 — Office hours (days 71-75 and ongoing weekly). Drop-in sessions where attorneys can ask questions, get help with specific scenarios, and provide feedback. Run these weekly for the first month, biweekly after that. Champion network: Identify 2-3 attorneys in each practice group who are natural early adopters. Give them early access during UAT, involve them in configuration decisions, and make them the first point of contact for their peers. Peer influence drives adoption faster than top-down mandates. Budget 15-20% of your implementation spend on change management. If you spent $30,000 on the tool and implementation, allocate $5,000-$6,000 for training materials, workshops, and the internal time of your champion network. This is the line item most departments skip — and it's the line item that determines whether the tool gets used or abandoned.

Days 76-90: Measurement and the Expansion Case

The final two weeks are about proving value and building the case for phase 2. Collect your metrics. Pull the data on your three success criteria defined on day 1. Compare before and after. If you reduced NDA review time from 3 days to 1.2 days, that's a 60% improvement. If you processed 67 contracts through the platform, you exceeded your 50-contract target. If first-pass approval rate is 93%, you beat the 90% target. Calculate dollar impact. Translate time savings into dollars. If 15 attorneys each save 3 hours/week through AI-assisted workflows, that's 45 hours/week x 48 weeks = 2,160 hours/year. At a $200/hour fully-loaded attorney cost, that's $432,000 in annual capacity created. Frame it as 'equivalent to adding 1 FTE without adding headcount.' Build the expansion proposal. Use your 90-day results to justify phase 2 — either expanding the current use case to more teams or adding a second AI tool for a different workflow. Include actual results (not projections), lessons learned, and a specific budget ask for the next 90 days. Document what didn't work. If adoption was lower than expected in certain practice groups, or if the tool struggled with specific contract types, document it honestly. Credibility with leadership comes from transparency, not cheerleading.

The Bottom Line: The 90-day legal ops AI playbook: spend days 1-15 defining the problem and success criteria, days 16-35 selecting from a shortlist of 3 vendors, days 36-60 implementing in phases with a single named owner, days 61-75 running a three-part training program with champion attorneys, and days 76-90 measuring results and building the expansion case. The departments that follow this sequence go from zero to measurable ROI in one quarter. The ones that skip steps spend a year getting nowhere.