I build AI into real businesses every week. These are the patterns I keep seeing: what works, what's theater, and where the money actually is. None of it is gospel. All of it is field-tested.
Every week an owner asks me where to start with AI. Wrong question. The right question is what to stop doing first. AI is step five of five. Most teams try to make it step one, which is how you end up automating a process that never deserved to exist.
Most process steps exist because someone added them years ago and nobody asked why. Every requirement needs an owner with a name, not a department.
The cheapest automation is a step that no longer exists. If you're not adding 10% of deleted steps back later, you didn't delete enough.
Collapse approvals, kill the duplicate data entry, standardize the inputs. Simple processes are the only ones worth speeding up.
Tighten handoffs and cycle time with the people you already have. Now the bottlenecks worth solving are visible.
Aim it where the tool does 80% of the work and a person owns the last 20%. Measure it in dollars and hours, never in demos.
This is why our diagnostics start in your operations, not in a model picker. The companies that skip steps one through four buy software. The ones that don't buy outcomes.
It's not "we use ChatGPT." Everyone uses ChatGPT.
AI-native means the workflow is designed assuming the tool does 80% of the work and your people supply the judgment. Not a tool bolted onto the old process; a new process built around the tool.
The test: if the AI disappeared tomorrow, would the process break? If the answer is no, you're AI-curious, not AI-native. Curiosity is fine. It's just not an advantage.
Read →Your best people already use AI. Quietly. Personally.
That's individual leverage, and it walks out the door at 5 o'clock. It lives in personal accounts, private prompts, and one person's head.
Institutional capability is what the business owns: systems in production, playbooks anyone can run, teams trained together. It compounds, survives turnover, and shows up in the valuation. Most "we're using AI" companies have the first and confuse it for the second.
Read →AI gets the meeting. Operations get the money.
Nobody actually wants AI. They want the close to take three days instead of three weeks, the AP clerk to stop keying invoices, the field crew trained without pulling them off a job.
AI enablement is the wedge. Ops modernization is the deliverable. If a proposal leads with models instead of your P&L, someone is selling you their product, not your outcome.
Read →Speed isn't a brag. It's how you keep scope honest.
A year-long engagement can hide anything. A two-week diagnostic can't. When the deadline is days away, every activity has to either find money or build something, and the theater falls away on its own.
That's the real reason we ship in weeks. Not because fast is impressive, but because short clocks make everyone tell the truth: about the data, the process, and whether the thing is working.
Read →15 minutes. No deck. No spam. Bring the process that annoys you most and I'll tell you straight whether AI belongs anywhere near it. And if you think one of these perspectives is wrong, even better. Bring that.