Where does AIactually fit?
The question we hear most from leaders isn't whether to use AI. It's where. This page is how we think it through with clients. Take it and use it.
How we look at the work
You don’t need to start with AI. Start with the work.
We follow one process from beginning to end, and open the work up to see what is actually happening. What we find decides what gets built.
Visible work
A process looks simple from the outside.
Under the work
Five layers run underneath: people, process, data, systems, rules.
Reorganized
Repeatable work goes to systems. Judgment stays with people.
The visible work
Simple from the outside.
Under the work
The hidden operating system
Reorganized
Human work
Judgment · context · relationships · creativity · accountability
System work
Repeatable · rules-based · standardized · automated
Foundation
Connected systems · clean data · clear rules · defined ownership
Now we know where technology belongs.
AI shows up three ways. Who's driving?
AI isn't one thing you adopt.
Where most teams start
Hand it off
The formal term: automation.
The recurring, rule-based work: same steps, same shape, every cycle. AI carries it.
Pulling the same numbers into the same monthly summary.
What stays human: Someone still owns the output.
Work it together
The formal term: augmentation.
The thinking work: drafting, analyzing, weighing options. The person steers the whole way.
A first draft in minutes that you shape into the real thing.
What stays human: Every call along the way.
Set it running
The formal term: agency.
AI acts on your behalf inside limits you set: clear checkpoints and review loops.
A request comes in, gets sorted and routed, and a person approves anything unusual.
What stays human: The guardrails, and the final say.
Most operations use all three. The skill is telling them apart, because each one is set up, checked, and trusted differently.
Fluency isn't technical. It's judgment, practiced.
Working well with AI doesn't require an engineering background. It comes down to four skills every experienced operator already recognizes.
01
Choose
Knowing what to hand off and what to keep. Not everything belongs with AI.
02
Ask
Saying what you actually need: the context, the constraints, what good looks like. The quality of what comes back is set by the quality of what goes in.
03
Judge
Reading what comes back with a practiced eye: what's right, what's off, what's missing.
04
Own
Staying responsible for what goes out the door. However the work got done, your name is on it, and your standards apply.
The thinking here is grounded in Anthropic's AI Fluency framework, which our team is certified in. These four words are how we apply it inside real operations.
Execution moves to systems. Judgment, relationships, and the final call stay with people.
What matters is a champion close to the work, and leadership using the tools before asking anyone else to. How the work is organized matters about twice as much as any one person's skill with it, and that's the part you control. The room people get back is the advantage.
The dip is normal. Plan for it.
Even well-resourced rollouts dip before they stick. What gets teams through it is unglamorous, a standing monthly review:
Findings drawn from Microsoft's 2026 Work Trend Index Annual Report and Microsoft's own rollout experience. It is vendor research, and we read it that way.
Could AI carry one of your tasks? Try it now.
Pick one task your team does every week and answer six questions about it. Nothing is recorded, and nothing leaves this page.
Does the task repeat on a rhythm, daily, weekly, or monthly?
Could you write down the rules for how it's done?
Do the inputs arrive in roughly the same shape each time?
Does it take meaningful team hours every week?
If it were done imperfectly on day one, would a person catch it before it mattered?
Would a person still review the result before it reaches a client, your board, or your books?
0 of 6 answered
Clarity first.Then the build.
The judgment on this page comes from the builds we've delivered. If you're carrying the question of where to start, that's the conversation we like most.