In this blog post I'm going to walk you through the exact 90-day roadmap I use to ship AI implementation engagements for founder-led businesses. Day-by-day in the first two weeks, week-by-week through the rest. Not the version that exists as a beautiful slide and never makes it to actual implementation. The version that produces 2-3 shipped AI workflows in 12 weeks, every time.
Most AI implementation roadmaps you'll find online are written by consultants who sell roadmaps. They're 47 pages long. They have phases with names like "Discover," "Define," "Develop," "Deploy" (the four Ds — every consulting deck has them). They look impressive. They rarely produce shipped work because they're designed to look thorough, not to actually ship.
The roadmap below is designed to ship. Six steps over 90 days. Each step has a specific deliverable. No "transformation phase" or "alignment workshop" or "stakeholder discovery sprint." Just shipping work.
I've been a marketing consultant for twenty-one years. I went all in on AI implementation work in 2024. The roadmap below is what I actually use, refined over 30+ implementation engagements since. My clients have included IBM, Twitter, Dropbox, monday.com and Greenpeace. I run a newsletter at 15,000 subscribers with a 70 per cent open rate.
By the end of this blog you'll have the exact 90-day roadmap, the deliverables at each milestone, the common ways it goes wrong, and how to use it whether you're hiring a consultant or DIY-ing.
TL;DR
Six steps over 90 days:
- Days 1-7: Diagnosis. Map current workflows, identify the 2-3 with highest AI leverage. Deliverable: prioritised workflow list.
- Days 8-21: Design. Detailed blueprint for the first workflow including tools, integrations, human checkpoints. Deliverable: workflow blueprint document.
- Days 22-45: Build #1. Construct and test the first workflow. Deliverable: working system on staging.
- Days 46-60: Ship #1, Design #2. First workflow goes live; second workflow blueprint complete. Deliverable: workflow #1 in production + workflow #2 design.
- Days 61-80: Build + Ship #2. Construct, test, ship the second workflow. Deliverable: workflow #2 in production.
- Days 81-90: Handoff. Documentation, team training, off-boarding plan. Deliverable: team owns the workflows, consultant exits.
What's NOT in this roadmap: strategy phase (already done before day 1), transformation phase (not needed for 2-workflow scope), executive alignment workshops (the founder is the alignment).
Why most roadmaps fail to ship
Three structural reasons most AI implementation roadmaps never produce shipped work:
Too much front-loaded discovery. Roadmaps that allocate 4-6 weeks to "discovery and strategy" before building anything are designed to invoice you for two months without producing anything. Real discovery for a focused 2-workflow engagement takes 1-2 weeks max. Anything more is padding.
Too many stakeholders. Roadmaps that require "alignment workshops" with 8-12 stakeholders are designed to slow shipping. For most founder-led businesses, the founder is the only stakeholder that matters strategically; the operational team can input but doesn't need to approve every decision.
No accountability for shipping. Roadmaps that describe phases without committing to specific shipped deliverables have no accountability. Real roadmaps say "by week 6, X is live in production." Vague roadmaps say "by week 6, we'll be in the build phase."
The roadmap below avoids all three.
The 90-day path, day by day for the first 2 weeks
Week 1: Diagnosis
Day 1 (3 hours): Discovery call with founder. Goal: map current marketing/business workflows at a high level. Output: list of 8-15 workflows currently running, with rough time-per-week estimate for each.
Day 2 (2 hours): Review with one operational team member who actually runs the workflows. Goal: validate the time estimates and catch missed workflows. Output: refined workflow list with reality-checked numbers.
Day 3-4 (4 hours total): Score each workflow on three criteria: (1) time consumed per week, (2) quality maintained or improved by AI, (3) implementation complexity. Output: prioritised list with top 2-3 candidates for AI implementation.
Day 5-6 (3 hours): Present prioritised list to founder. Validate the top 2-3. Confirm engagement scope. Output: signed-off scope document covering workflows 1 and 2.
Day 7: Buffer day. (Things take longer than planned in week 1.)
Week 1 deliverable: Signed-off scope document committing to ship 2 specific workflows by day 90.
Week 2: Design workflow #1
Day 8-9 (5 hours): Deep map of the current state of workflow #1. Every step. Every input. Every output. Every human decision point. Document the actual current process, not the assumed one.
Day 10-11 (4 hours): Design the AI-assisted version. Tool selection, integration points, where humans stay in the loop, what the AI inputs look like, what the AI outputs look like, what happens when the AI gets it wrong.
Day 12-13 (3 hours): Review design with the team member who runs the workflow currently. Iterate based on their input. Document failure modes they've seen before.
Day 14 (2 hours): Final design document. Includes: tools selected, prompts (if applicable), integration architecture, human-review checkpoints, failure modes and handling, expected time saved.
Week 2 deliverable: Workflow blueprint document for workflow #1.
Weeks 3-6: Build workflow #1
Week 3: Tool setup and account provisioning. Week 4: Build the workflow on staging/test environment. Week 5: Test with real example work. Week 6: Iterate based on testing. Polish prompts, fix edge cases.
Week 6 deliverable: Workflow #1 working end-to-end on staging. Tested with at least 5 real business examples.
Weeks 7-9: Ship workflow #1, design workflow #2
Week 7: Bring workflow #1 live in production. Run for the first time on actual current business work. Founder reviews output. Calibrate.
Week 8: Continue running workflow #1 in production. Track time savings vs old method. Document edge cases that emerge.
Week 9: Workflow #1 stable in production. Begin design for workflow #2 (same process as weeks 2 for #1).
Week 9 deliverables: Workflow #1 in production (live for 2+ weeks) + workflow #2 design complete.
Weeks 10-12: Build, ship, hand off workflow #2
Week 10: Build workflow #2.
Week 11: Ship workflow #2. Test in production.
Week 12: Documentation for both workflows. Training session with the team. Off-boarding plan. Final review with founder.
Week 12 deliverables: Both workflows in production, both documented, team trained to operate both without the consultant.
What each deliverable actually looks like
Workflow blueprint document (1-2 pages per workflow): - Current state map - AI-assisted future state - Tools used + cost - Integration points (CRM, email, etc.) - Human review checkpoints - Expected time saved - Failure modes and handling - Success metrics
Working system (delivered as a runnable workflow): - Account access set up - Prompts saved and versioned - Integrations functioning - Test results documented - Edge cases catalogued
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Team documentation (1-page operator manual per workflow): - How to run the workflow step by step - What to do when the AI gets it wrong - Where to find the prompts and edit them - Who to contact when something breaks (the team member who owns the workflow, not the consultant) - Monthly review checklist
Off-boarding handover (final meeting): - Demo both workflows live with the team - Q&A - 30-day post-engagement support arrangement (optional, separately scoped)
Common ways this roadmap goes wrong (and how to fix)
The diagnosis phase produces 8 priority workflows instead of 2-3. Cause: too many stakeholders inputting. Fix: founder makes the final call, not the team.
Week 2 design takes 4 weeks because of stakeholder reviews. Cause: too many sign-off steps. Fix: design with 1-2 operational people, not 5+.
Workflow #1 doesn't ship in week 7 because tools weren't approved. Cause: didn't get IT/admin approval upfront. Fix: get tool approvals during week 1 alongside the scope sign-off.
Workflow #1 ships but team doesn't use it because they prefer the old way. Cause: team wasn't involved in the design. Fix: the operational team member who runs the workflow is co-designer, not just consulted.
Workflow #2 doesn't ship because workflow #1 needed more iteration. Cause: underestimated the complexity of workflow #1. Fix: pick less ambitious workflows for the first engagement; build complexity over time.
Engagement ends but team can't operate workflows independently. Cause: documentation was an afterthought. Fix: documentation is a week 11-12 deliverable, not a week 12 add-on.
How to use this roadmap
If you're hiring a consultant: Send this roadmap to the consultant and ask which steps they'd compress or expand and why. Their answer reveals how they think about implementation. Consultants who push for more discovery time are usually padding. Consultants who immediately want to compress the design phase often skip the failure-mode work that prevents production issues.
If you're DIY-ing:
If you're managing a junior consultant: This is your accountability framework. Specific deliverables at specific weeks. If they miss week 6 (workflow #1 built on staging), the engagement is in trouble. Don't accept "we're still in the build phase" past week 7.
What I do for clients
If you want to talk about your specific 90-day path, the discovery call is the place.
Frequently asked questions
Can this roadmap work in 60 days instead of 90?
Possibly, for a 1-workflow engagement. Not for 2 workflows. The compression usually means skipping the documentation phase, which is the bit that prevents future failure.
What if my business needs 5 AI workflows, not 2-3?
Run two consecutive 90-day engagements. Workflows 1-2 in days 1-90, workflows 3-4 in days 91-180. The compounding learning from the first engagement makes the second one faster. Trying to ship 5 workflows in one 90-day engagement usually produces 5 mediocre implementations.
Can the roadmap start with strategy if we don't have one yet?
What if our team can't dedicate 3-5 hours per week to the engagement?
The roadmap doesn't work. AI implementation requires team participation. If the team is too busy to participate, hire the implementation as fully done-for-you with no team involvement (cost: 2-3x the standard engagement), or wait until the team has capacity.
Do I have to use this specific roadmap?
No. This is the one I use because it works for my engagement type. Other consultants have their own frameworks that work for them. The principles that should be in any roadmap: short discovery, fast first ship (week 6 maximum), team participation, off-boarding by week 12.
How do you decide which workflow to ship first?
Three criteria, in order: (1) highest weekly time savings, (2) clearest before/after metric, (3) lowest implementation complexity. The intersection of "high time savings" and "low complexity" is where to start. Save the complex workflows for later when the team has built confidence with the first one.
What happens after the 90 days?
Three options. (A) End the engagement, team runs the workflows independently. (B) Reduced retainer (1-3 days per month) to identify the next workflow and provide advisory. (C) Full second engagement to ship workflows 3-4. Most clients pick B for 3-6 months then move to occasional advisory.
Can the roadmap deliver compound results, or is it one-time?
The first 2 workflows are the foundation. They compound by freeing up team time AND by teaching the team how AI implementation works. By the third or fourth workflow, the team is implementing alongside the consultant. By the fifth workflow, the team is implementing on their own with consultant advisory. That's the compound.
The thing to take away
AI implementation roadmaps that look impressive on slides usually fail to ship. Roadmaps that ship are shorter, more specific, and built around shipped deliverables at specific weeks.
The 90-day roadmap above produces 2-3 working AI workflows in 12 weeks. The math: 6 weeks of design and build for the first workflow, 4 weeks for the second (faster because of the first), 2 weeks for handoff and documentation.
If your roadmap doesn't ship anything in production by week 7, it's the wrong roadmap.
If you want help running this roadmap for your business, the discovery call is twenty minutes and free.
Internal links: - /work-with-me/ - /about/ - /what-is-ai-implementation/ - /how-much-does-an-ai-consultant-cost-2026/ - /ai-marketing-consultant-2026/ - /ai-for-service-businesses/
Schema: Article + Person @id + FAQPage + HowTo + BreadcrumbList
Related from this week
- /ai-marketing-audit-framework/ — the 47-check pre-implementation audit
- /ai-for-seo-without-penalty/ — the AI-safe SEO playbook
- /rebuilt-marketing-business-with-ai/ — the case study of how I rebuilt mine
Related: How Much Does an AI Consultant Cost in 2026? Real Price Band
