The 90-Day AI Plan: How to Stop Debating and Start Moving

The 90-Day AI Plan: How to Stop Debating and Start Moving

Your organization has been discussing AI for 12 months. You've attended three conferences. You've commissioned two strategy documents. Your CMO wants to invest in content AI. Your CFO wants to audit AI spend. Your CTO wants to build a center of excellence. The CEO wants board-level reporting on AI readiness.

Meanwhile, nothing tangible has happened.

This is the consensus trap. Organizations stuck in endless debate about AI strategy aren't being thoughtful. They're avoiding the hard work of actually moving. The cure is a structured 90-day plan that forces accountability, creates momentum, and delivers measurable progress in one quarter.

Dr. Mark van Rijmenam, world-leading futurist and AI expert, developed the Intelligence Age Scorecard to help organizations see where they actually stand. What the Scorecard reveals is that waiting for perfect strategy is the oldest excuse for inaction. The organizations pulling ahead are the ones moving deliberately but fast, measuring ruthlessly, and iterating relentlessly.

Here's the 90-day plan that breaks the deadlock.

Why AI Debates Go in Circles

Debate loops form because the questions seem too big to answer without more information.

Should we build or buy? We don't know until we've seen what buying looks like. Should we use generative AI or traditional models? We don't know until we've tested both. Should we invest in reskilling or hire new talent? We don't know until we know what we're building.

Every question points to another question. The result: a year of conversations and no movement.

The way out isn't more deliberation. It's bounded action. You commit to a 90-day cycle where you:

1. Map the current state (what AI are we already using, where are we weak)
2. Set governance baseline (who decides, who approves, who measures)
3. Launch two pilots (pick the two highest-impact use cases that can move in 60 days)
4. Embed measurement (set KPIs, track them weekly, report to the board monthly)
5. Plan cycle two (learn from pilots, scale what works, kill what doesn't)

You're not solving for perfection in 90 days. You're solving for movement, learning, and accountability. That's how you escape the debate trap.

The 90-Day Plan Structure

Break the quarter into three 30-day phases, each with a specific output that becomes the input for the next phase.

Phase 1 (Days 1-30): Audit and Baseline Phase 2 (Days 31-60): Pilots and Learning Phase 3 (Days 61-90): Measurement and Board Motion

Each phase has a single owner—a member of your executive team who is accountable for delivery. This isn't a committee. It's a leader with P&L responsibility who gets measured on whether the phase ships.

Days 1-30: Quick Wins, Audits, Baselines

Your first month does three things simultaneously:

Audit existing AI usage. You think you don't have much AI yet. You're wrong. Start-ups and teams have probably already implemented ChatGPT, Copilot, or similar tools. They're using AI for scheduling, drafting, research, analysis. Your first job is to see what's already happening, where it's working, and where it's unsanctioned.

Schedule interviews with functional leaders (marketing, sales, ops, finance, HR). Ask: What are you already using AI for? What's working? What's blocked? What would you do with AI if governance and budget weren't constraints?

This audit serves two purposes. First, it gives you a realistic baseline of AI maturity. Second, it surfaces use cases that are already working and people who are already bought in. These become your pilots.

Establish governance baseline. You don't need perfect governance on day 30. You need workable governance. Here's the minimum:

• Who approves new AI projects (probably a steering committee with reps from tech, risk, ops, and the sponsoring function)
• How projects are evaluated (what criteria matter: ROI, risk, talent required, timeline)
• How data flows (what data can be used for training, what's off-limits)
• What gets measured (every pilot has three KPIs: adoption, impact, and risk)

Make these decisions in week two. Document them in week three. Get sign-off in week four. You're not designing a perfect governance model. You're designing one that lets you move.

Identify and scope two pilots. From your audit, pick two high-impact use cases. Pick one with high urgency (something that will show results in 60 days and prove value to skeptics) and one with high strategic importance (something that, if it works, reshapes how your organization operates).

Example pilot one: AI-assisted contract review in legal. High urgency, measurable output (hours saved per contract), pilot-able in 60 days.

Example pilot two: AI-driven demand forecasting in operations or supply chain. Higher complexity, longer horizon for full impact, but strategically important if you want to reshape planning cycles.

Define success criteria for each pilot before you start. What does success look like? How will you know it worked? Set a kill criterion: if we're not seeing signal by day 50, we pause and pivot.

Days 31-60: Pilots, Cross-Functional Teams

Your second month runs the pilots in parallel with the governance structure from month one.

Assemble cross-functional teams. Each pilot gets a team: the functional leader (the one with the problem), the CTO or relevant tech leader (the one who builds), and a measurement owner (the one who tracks KPIs). Teams meet weekly. This is not a slow-moving workstream. This is a focused group moving fast.

Build ruthlessly. Use off-the-shelf tools wherever possible. OpenAI, Claude, commercial AI platforms, no-code AI tools. The goal isn't to build cutting-edge models. The goal is to prove value. You can optimize later.

For contract review: set up a workflow where incoming contracts get fed to an AI service, which summarizes risks, flags deviations from template, and routes for human review. Measure how many hours lawyers save per contract and how much faster turnaround gets.

For demand forecasting: pull three years of historical data, train a model (or use a commercial service), run it against recent data to backtest accuracy, then start making live predictions. Track whether the new forecast reduces forecast error by at least 15%.

Communicate progress. Weekly updates to the steering committee. These are 15-minute check-ins: Here's what we learned this week. Here's where we're stuck. Here's what we're changing. This keeps executives in the loop and surfaces blockers early.

Test kill criteria at day 50. On day 50, pause and ask: Are we seeing the signal we expected? For the contract review pilot, are lawyers actually using the tool and saving time? For demand forecasting, is the model materially more accurate than the baseline?

If yes, keep running. If no, don't throw more resources at it. Kill it, learn from it, and plan a different pilot for cycle two.

Days 61-90: KPIs, Board Reporting, Cycles

Your third month embeds the measurement and planning that makes this repeatable.

Formalize KPIs. By day 61, you know what worked and what didn't. For each active pilot, define the operational KPIs that matter:

• How many users are adopting the tool daily or weekly
• What is the measurable impact (hours saved, accuracy improvement, cost reduction, quality lift)
• What is the business ROI (what's the annualized value if this scales)
• What is the risk score (data quality, model drift, compliance issues)

Track these on a dashboard. Update weekly. This becomes your source of truth for AI's value in your organization.

Present to the board. Your board wants to see progress. Give them clarity. Here's what we found in the audit. Here's what pilots are running. Here's what worked, what didn't, and what we're learning. Here's our plan for cycle two.

This isn't about board approval for every decision. It's about board transparency. They need to see that AI investments are being managed with discipline.

Plan cycle two. In your last week, the steering committee designs the next 90 days. Based on what worked:

• Scale the pilots that succeeded (move them from 50 users to 500, from one team to five)
• Launch two new pilots (informed by learning from cycle one)
• Address the governance gaps you discovered (did we underestimate data quality issues? Do we need more legal guardrails?)
• Adjust the measurement framework (do we have the right KPIs? Are we tracking the right things?)

How the Plan Personalizes to Your Weakest Areas

The 90-day plan is a structure. You fill it in based on where your organization is weak.

If you're weak on data quality, your pilots focus on use cases where data is already clean (customer data, transaction data) and you learn to fix data problems in parallel.

If you're weak on talent, your pilots emphasize tools that complement existing teams (AI-assisted work) rather than tools that require new skills.

If you're weak on governance, your steering committee is stronger. You add a Chief Risk Officer or General Counsel to increase governance rigor.

If you're weak on business alignment, you pick pilots that directly reduce costs or increase revenue. You need quick wins to build credibility.

The structure is fixed. The content is yours.

Take the Intelligence Age Scorecard

The 90-day plan works because it forces the organization to see where you actually stand instead of where you think you stand. Your audit reveals which functions are ready and which aren't. Your pilots reveal which use cases are real and which are theoretical.

The Intelligence Age Scorecard does the same thing faster. In 15 minutes, you see your readiness across technology, skills, data, organization, governance, and business use cases. You see where you're strong and where you're weak.

Take the assessment today at thedigitalspeaker.com/intelligence-age-scorecard/, then use what you learn to inform your 90-day plan. The organizations moving fastest are the ones that measure where they stand, move deliberately, and iterate relentlessly.

Your 90 days start now.