ChatGPT
Google AI
Claude
Perplexity
Grok
Why 90 Days?
Ninety days is long enough to build proper foundations and see measurable results, but short enough to maintain urgency and executive attention. Companies that try to plan for 12 months before taking action often stall. Companies that rush into AI without governance create risks.
This 90-day roadmap strikes the balance: structured enough to manage risk, fast enough to capture value.
Before You Start: Pre-Requisites
Before beginning the 90-day roadmap, ensure you have:
- Executive sponsorship — At least one C-suite sponsor who will champion the programme
- A small governance team — 3-5 people from IT, legal/compliance, HR, and business operations
- Budget clarity — Know what you can spend on tools, training, and external support
- Baseline assessment — A rough understanding of where AI is already being used (formally or informally)
Phase 1: Foundation (Days 1-30)
Week 1: Assess and Align
Objective: Understand current state and set direction.
- Conduct an AI readiness survey across the organisation (anonymous, 10-15 questions)
- Interview 5-10 department heads about their AI pain points and aspirations
- Audit existing AI tool usage (what tools are employees already using?)
- Review competitor AI adoption (what are peers in your industry doing?)
- Align with executive sponsor on goals, budget, and success metrics
Deliverable: AI Readiness Report — current state, gaps, and opportunities
Week 2: Governance Setup
Objective: Establish the governance framework.
- Form the AI Governance Committee (3-5 members, cross-functional)
- Draft the company AI policy (use our AI Policy Template)
- Draft the AI acceptable use policy for employees
- Define the AI tool approval process and checklist
- Identify the Data Protection Officer's role in AI governance
Deliverable: Draft AI Policy and Acceptable Use Policy
Week 3: Tool Selection and Approval
Objective: Approve initial AI tools for company use.
- Evaluate 2-3 enterprise AI tools against your approval checklist
- Negotiate enterprise licences with preferred vendors
- Set up enterprise accounts with SSO, admin controls, and audit logging
- Configure data loss prevention (DLP) rules for AI tools
- Block unapproved AI tools at the network/proxy level (if feasible)
Deliverable: Approved tools list and enterprise accounts configured
Week 4: Training Design
Objective: Plan the training programme.
- Identify training cohorts (who gets trained first?)
- Select a training provider or design internal training
- Schedule training sessions for Phase 2
- Create role-specific use case libraries
- Apply for HRDF (Malaysia) or SSG/SFEC (Singapore) funding
Deliverable: Training schedule and funding applications submitted
Phase 2: Activate (Days 31-60)
Week 5-6: Training Rollout
Objective: Build AI skills across the organisation.
- Deliver AI training to the first cohort (AI champions / early adopters)
- Distribute the AI acceptable use policy to all employees
- Launch the prompt library on your intranet or shared drive
- Set up an internal AI help channel (Slack/Teams) for questions
- Collect feedback from the first training cohort and adjust
Deliverable: First cohort trained, AUP distributed, internal support channel live
Week 7-8: Pilot Projects
Objective: Demonstrate value with quick wins.
- Select 3-5 pilot AI projects across different departments
- Each pilot should have a clear problem, an AI solution, and a success metric
- Assign an AI champion to lead each pilot
- Document the process, results, and lessons learned for each pilot
- Present pilot results to the executive sponsor and governance committee
Example pilot projects:
| Department | Pilot Project | Success Metric |
|---|---|---|
| Marketing | AI-assisted content creation | Time saved per content piece |
| Finance | AI-powered report summarisation | Hours saved per week |
| HR | AI-assisted job description writing | Time to publish reduced |
| Customer Service | AI response drafts for common queries | Response time reduction |
| Operations | AI analysis of process data | Insights generated per week |
Deliverable: 3-5 pilot projects completed with documented results
Phase 3: Scale (Days 61-90)
Week 9-10: Expand Training
Objective: Train the broader organisation.
- Deliver AI training to remaining departments/cohorts
- Offer advanced training to AI champions
- Update training content based on pilot learnings
- Create department-specific prompt libraries based on pilot successes
- Launch an internal AI newsletter or knowledge-sharing session
Deliverable: Organisation-wide training complete
Week 11-12: Institutionalise
Objective: Make AI part of normal operations.
- Finalise and formally publish the AI policy (CEO endorsement)
- Embed AI governance into existing risk management processes
- Set up quarterly AI tool reviews and policy updates
- Define ongoing AI training requirements (annual refresher)
- Create an AI use case backlog for future projects
- Present the 90-day results to leadership, including ROI data
- Plan the next phase of AI adoption (months 4-12)
Deliverable: Formal AI policy published, governance embedded, leadership report
Success Metrics
Track these metrics throughout the 90 days:
| Metric | Target | How to Measure |
|---|---|---|
| Employees trained | 80%+ of target cohort | Training attendance records |
| AI policy awareness | 90%+ of employees | Survey or policy acknowledgement |
| Approved tools adoption | 50%+ of trained employees actively using tools | Tool usage analytics |
| Pilot projects completed | 3-5 | Project completion tracker |
| Time/cost savings from pilots | Measurable improvement | Department-reported metrics |
| AI incidents | Zero critical incidents | Incident log |
Common Pitfalls to Avoid
- Starting with technology instead of governance — Approving tools before having policies leads to uncontrolled risk
- Training without follow-up — Skills decay quickly without reinforcement and ongoing support
- Piloting without measurement — If you do not measure results, you cannot justify scaling
- Excluding legal/compliance — AI governance is not just an IT responsibility
- Moving too slowly — If your 90-day plan takes 6 months, you lose momentum and executive attention
Adapting for Your Company Size
| Company Size | Adjustments |
|---|---|
| Small (10-50 employees) | Combine governance committee into 2-3 people; train everyone at once; 1-2 pilot projects |
| Medium (50-200 employees) | Follow the roadmap as written |
| Large (200+ employees) | Add a dedicated AI programme manager; train by department over 4-6 weeks; 5-10 pilot projects |
Related Reading
- AI Champions Program — Build the internal team that drives your roadmap forward
- AI Training for Executives — Executive buy-in is essential for your adoption roadmap
- 2-Day AI Workshop — Kick off your roadmap with a comprehensive team workshop
Sustaining Momentum Beyond the First 90 Days
The 90-day roadmap establishes foundations, but lasting AI adoption requires a structured transition into ongoing operations. At the conclusion of the initial roadmap, organizations should formalize the AI governance structure, transition pilot projects into production deployments with defined support models, and establish a quarterly AI strategy review cadence. Creating an internal AI community of practice where early adopters mentor later adopters extends the roadmap's impact beyond the initial champion group. Organizations that fail to plan beyond 90 days frequently experience adoption plateaus where initial enthusiasm fades without sustained organizational support structures.
Beyond the structured roadmap phases, organizations should track leading indicators that predict long-term adoption success. Early warning signs of adoption stalls include declining tool login frequency after initial training peaks, increasing support ticket volumes for basic features that training should have covered, and manager disengagement from progress reviews. Addressing these signals proactively through targeted refresher training, workflow redesign, or additional change management support prevents the common pattern where initial enthusiasm fails to translate into sustained organizational capability.
Common Pitfalls That Derail 90-Day AI Adoption Plans
The most frequent reason adoption roadmaps stall is scope creep during the pilot phase, where initial use case selections expand to include adjacent processes before the original pilot demonstrates clear value. Another common pitfall is underestimating change management requirements, particularly in organizations where employees have experienced previous technology rollouts that disrupted established workflows without delivering promised benefits. Setting realistic expectations during the roadmap kickoff and celebrating incremental wins throughout the 90-day period helps maintain organizational momentum.
Common Questions
Yes. A 90-day roadmap will not transform the entire organisation, but it will establish governance, train key teams, complete pilot projects, and demonstrate measurable value. This creates the foundation and momentum for ongoing AI adoption in months 4-12 and beyond.
Governance should come first, but only slightly. In the 90-day roadmap, governance setup happens in weeks 2-3, and training starts in week 5. You need basic policies and approved tools in place before training employees, but do not let governance work delay training by more than 2-3 weeks.
Costs vary based on company size and ambition. For a mid-size company (50-200 employees), budget for: enterprise AI tool licences (S$10-30 per user/month), training (S$5,000-25,000 for workshops, often subsidised), and optionally external consulting support. Total investment is typically S$20,000-80,000, much of which is reclaimable through government subsidies.
Ideally a senior business leader (VP or Director level) with cross-functional credibility, executive sponsorship, and 25-30% time allocation. This person chairs the AI governance committee and owns the roadmap delivery. Smaller companies may assign this to CTO/COO; larger companies may hire a dedicated AI Programme Manager.
Executive sponsorship is non-negotiable. Without C-suite backing, you'll face budget blocks, political resistance, and low priority when conflicts arise. If you lack sponsorship, build the business case first: show competitor moves, ROI projections, and regulatory/market pressures. Present to executives before launching the roadmap.
It depends on internal capabilities. Consultants add value for: governance framework setup (legal/policy expertise), training delivery (accelerates quality), pilot project design (industry-specific use cases), and change management. DIY works if you have strong IT, willing executives, and someone with 50%+ time to drive it.
The 90-day plan creates foundations—governance, trained teams, proven use cases. After Day 90, focus on: scaling pilots company-wide, expanding training to remaining departments, refining policies based on learnings, building advanced use cases, and establishing ongoing governance (quarterly reviews). Plan for continuous improvement over 12-24 months.
References
- AI Risk Management Framework (AI RMF 1.0). National Institute of Standards and Technology (NIST) (2023). View source
- ISO/IEC 42001:2023 — Artificial Intelligence Management System. International Organization for Standardization (2023). View source
- Model AI Governance Framework (Second Edition). PDPC and IMDA Singapore (2020). View source
- Enterprise Development Grant (EDG) — Enterprise Singapore. Enterprise Singapore (2024). View source
- What is AI Verify — AI Verify Foundation. AI Verify Foundation (2023). View source
- OECD Principles on Artificial Intelligence. OECD (2019). View source
- Training Subsidies for Employers — SkillsFuture for Business. SkillsFuture Singapore (2024). View source