The First 90 Days: Foundations
The first 90 days are about learning the ground, picking the right tools, and getting every person started.
The first 90 days set the shape of everything that follows. The goal is not to deploy AI across the whole organization. It's to understand the ground, pick a small set of tools, and get every person started on a real learning path.
Four things happen in parallel.
Assess what you have
Select and set up tools
Train the first wave
Build Relationships
Assess what you have
Before picking tools, audit the workflows and systems you already run. You are looking for two things: where people lose time today, and where AI could plausibly help without adding risk.
A simple audit loop works well.
You repeat this for each team. It takes longer than you expect. It's also where most of the real insight comes from, because the gap between how a tool is supposed to be used and how it's actually used is almost always a flag for finding good AI use cases: the smaller the gap, the easier it will be to implement AI.
AI is an amplifier. A well defined SOP can be empowered reliably with AI, and poorly defined one will go wrong more quickly.
Select and set up tools
Pick a small number of tools. For most small and medium organizations, that means one general-purpose AI assistant plus any domain-specific tools the work genuinely needs.
Set them up properly. That means real accounts (not personal ones), single sign-on if you have it, clear data rules about what can go into which tool, and a documented owner for each system.
Train the first wave
Every staff member gets a baseline session and a personal learning plan tied to their actual job. An accountant's plan should not look like a coach's plan.
The Axios job listing for this exact role puts it well: meet teams where they are, and provide tailored support based on their level of familiarity and confidence Greenhouse. A one-size training ends up fitting no one.
Build relationships
Spend real time with the people whose workflows you are about to change. In this case that means coaches. You cannot prescribe AI use for work you don't understand, and the people doing the work know things you don't.
A worked example
Here's what this can look like in a workplace already using Claude Cowork, Monday.com, and the Microsoft Office suite including Teams.
Week 1 to 2: audit. You sit with two people from each team and watch them work. You notice the operations team spends several hours every Friday pulling project data out of Monday.com and writing status updates that get posted in Teams. Coaches keep their session notes in Word documents across scattered OneDrive folders. Nobody can find anything from last quarter.
Week 3 to 4: pick the first wins. You define two concrete use cases. First, a weekly status summary: Cowork reads the relevant Monday.com board and a set of Teams channels, then drafts the Friday update in a Word template. Second, notes cleanup: Cowork takes a folder of coach session notes and produces a structured summary by scholar. Both use Cowork's ability to read local files and move between applications without the person coordinating each step.
Week 5 to 6: set up and document. You configure Cowork with access to the specific OneDrive folders and Monday.com board that matter. You write a one-page runbook for each use case: what it does, what data it touches, who owns it, and what to do if it breaks. You set data rules, for example that scholar names are fine in Cowork but financial account numbers are not.
Week 7 to 10: train in waves. Operations team first, because they have the clearest problem and the most to gain. They see the Friday summary get drafted in ten minutes instead of three hours. Word spreads. Coaches come next, with a session focused on their specific notes workflow. Every staff member leaves with a personal learning plan: one thing to try this week, one thing to try next month.
Week 11 to 12: measure and hand off. You track two things: how many staff members are actively using the tools, and how much time the two use cases are actually saving. You write the first version of the strategy note for the CEO. You make sure both workflows could be picked up by someone else if you disappeared tomorrow.
What good looks like at day 90
You have finished a real assessment of the tech stack and team-by-team readiness.
Tools are selected, configured, and documented.
Every staff member has had a training session and a personal learning plan.
You have built working relationships with every team.
