The first 90 days of an SMB AI transformation, from someone who's run them
No transformation office, no seven-figure consulting engagement. Here's the 90-day sequence that actually works at a 20 to 500 person company, including the two weeks where everyone wants to quit.
In short
An SMB AI transformation doesn't need a transformation office. It needs 90 disciplined days focused on one function at a time.
- Weeks 1 to 2: pick one function with measurable output and baseline how the work actually happens.
- Weeks 3 to 6: train that team hands-on on their own tasks, with simple data guardrails.
- Weeks 7 to 10: rebuild two or three core workflows AI-first and retire the copy-paste versions.
- Weeks 11 to 13: measure hours saved, publish wins internally, and seed the next function.
- The most common failure is starting with tools instead of skills; the second is starting everywhere at once.
Decide what you're not doing
An SMB AI transformation succeeds or fails on what you decline to do. Most die in the first week, at the scoping meeting, when 'let's bring AI to the company' turns into eleven simultaneous initiatives. You don't have the management bandwidth for eleven. You have bandwidth for one function done completely, and that constraint is an advantage: it forces depth, and depth is where the returns are.
Pick the function where output is measurable and the pain is loud: usually sales, support, or finance. Pick a leader there with genuine curiosity about the tools; someone who is merely compliant will stall you by week four. And set the success metric on day one, in operational units: hours saved per person per week, deals touched, tickets resolved. Never 'adoption.'
Having run this at enterprise scale and at startup scale, I'll take the small company every time. What you lack in enablement budget you make up in speed: a decision that would need a committee at an enterprise takes one conversation here.
The 90-day SMB AI transformation sequence, phase by phase
The sequence runs in four phases over 90 days, one function at a time. Skills come before workflow rebuilds, and measurement comes before you expand.
- 1
Weeks 1-2: baseline reality
Sit with the team and map how the work actually happens, which is rarely what the process doc says. List every task by frequency and pain. Set guardrails: what data is safe for AI, what never leaves.
- 2
Weeks 3-6: skills, not tools
Train the team hands-on, each person on their own tasks. The goal is reps on real work until AI-first stops taking willpower. This phase decides everything that follows.
- 3
Weeks 7-10: rebuild workflows
Take the two or three highest-frequency workflows and redesign them AI-first, end to end. Retire the old version, or people will quietly revert under deadline pressure.
- 4
Weeks 11-13: measure and expand
Count hours saved and work shipped. Publish the wins with names attached. Then take your two strongest operators and seed the next function with them.
Weeks 5 and 6 are where it almost dies
Every rollout has a trough. Around week five, novelty has worn off, deadlines are loud, and the new way still costs a little more effort than the old way, because nobody has crossed the adoption inflection point yet. This is when teams quietly revert, and dashboards won't tell you until a month later.
Two things carry you through the trough. A coach available at the exact moment someone gets stuck, because stuck-and-alone is how reverting starts. And a leader who keeps doing their own work AI-first in public, because teams copy what leaders do, not what they announce.
Resist adding tools during the trough. The fix for friction is never another subscription; it's more reps on the tools you have. Tool-switching at week five is the most expensive form of procrastination I know.
Set these before you start
Common questions
How long does an SMB AI transformation take?
The first function takes about 90 days to show measurable results: two weeks of baselining, four of hands-on training, four of workflow rebuilds, and three of measurement. Company-wide is a sequence of those, accelerating as internal champions multiply.
What should an SMB transform with AI first?
The function with measurable output and loud pain, usually sales, support, or finance. Depth in one function beats breadth across all of them, because depth produces the proof and the champions that pull everyone else along.
Do we need consultants for an AI transformation?
At SMB scale, usually not. You need skills in the team that owns the work, simple guardrails, and a disciplined sequence. That's exactly what Candova AI's team training provides, with Cando coaching each person on their actual job, at a fraction of a consulting engagement.
Run your first 90 days with us
We'll map the sequence to your team and train them on their real work.
Power users save 10+ hours a week. Learn how.
The practical AI habits behind it, one a week.
Written by
Adrián Ridner
Co-founder of Candova, founder of Study.com, and O'Reilly AI author
Adrián has spent two decades as a serial entrepreneur opening the doors to the life-changing impact of education. Before Candova, he founded and scaled Study.com into the largest platform for online college-credit courses, certification prep, and career-aligned degree pathways, helping millions of learners earn credentials for the modern workforce.