Most growing companies should buy almost all of their AI and build almost none of it. The only thing worth building is the piece tied directly to why customers choose you. Everything else is a commodity, and paying to custom-build a commodity is one of the most expensive mistakes I see founders make right now.
The 1995 problem
In 1995, every business owner was hearing the same thing: you need to be on the internet. Some ignored it and spent the next decade catching up. Others overcorrected and spent fortunes building technology that was about to be free.
The winners did something simpler. They bought what everyone else was buying, a website, email and a way to take payments, and they put their own money and energy into what made them different.
AI is at the same moment. Doing nothing is a decision, and the companies getting this right now will be very hard to catch in five years. But the opposite mistake is just as real. I keep meeting founders who are paying developers to build custom chat assistants, custom note-takers and custom reporting tools that they could rent tomorrow for a monthly fee.
Why "more AI" isn't the goal
There's a quiet belief in a lot of leadership teams that more AI in more places equals progress. Every department should have something. The board wants to hear about initiatives.
Coverage is not the goal. Return is.
What you actually want is the smallest set of AI-enabled roles that meaningfully changes your economics. Three that work beat twelve that are half-running. Each one you add takes attention, data, training and oversight, and none of that is free.
The four filters
Before an AI idea gets budget, run it through four questions. Miss one, and it waits.
1. Value: is there money or capacity at stake? If this worked perfectly, would revenue, margin or the customer experience change in a way you'd notice? "It would be nice" doesn't count.
2. Repetition: is there enough recurring work? AI pays off through volume. A task that happens twice a quarter isn't worth automating. A task that happens 200 times a week probably is.
3. Fit: can AI do this work credibly? Some work is a great fit today: drafting, summarizing, sorting, extracting data from documents. Some work still needs a person's judgment. Be honest about which is which.
4. Readiness: can we actually run it? Do you have clean data, a written process, someone to own it and a way to check its work? If not, the idea isn't bad. It just isn't ready.
Sort into four piles
Once you've run your ideas through the filters, each one lands in one of four piles:
- Deploy or pilot. High value, good fit, tied to a real priority. Start here.
- Enable first. Useful, but blocked by data, process or ownership. Fix the blocker before you spend on the tool.
- Commodity: buy it. Useful but not what makes you different. Rent it off the shelf.
- Ignore for now. Interesting, but not connected to an outcome you care about.
The "ignore for now" pile will be bigger than you expect. That's the filter doing its job.
Buy the commodity, build the advantage
Here's the simple test for anything in the build conversation: if a competitor could buy the same capability next week, it's a commodity. Rent it.
Build only where your advantage lives. That's usually one of three places: a process you run better than anyone in your market, data that only you have, or judgment your best people carry that you want applied more widely. That's where custom work compounds, because nobody can buy their way to the same result.
Whatever you build, lead with security. Set the access rules and the data rules before a line of code gets written. At the end of the day a custom tool without guardrails is a liability with a nice interface.
What to do this week
Ask your leadership team to list every AI idea on the table: the pilots running, the vendors pitching, the things people keep bringing up. Put them all on one page.
Then go line by line and ask two questions out loud: "Could a competitor buy this?" and "Which of the four filters does it fail?"
Anything a competitor could buy goes in the buy pile. Anything that fails a filter goes in enable first or ignore for now. What's left is your real list. It'll be short.
AI Opportunity Map
The AI Opportunity Map walks you through five steps: your business context, where the friction is, what AI you already have in place, where the opportunities are, and a prioritized map of which AI roles to deploy, enable first, buy or ignore. Bring your whole list of AI ideas, including the ones you're already paying for, and let the four filters sort them. Be honest on readiness. That's the filter most people want to skip.
Open the AI Opportunity Map →Get the next lesson in your inbox
Short, practical lessons like this one for founders running $1M–$20M companies, plus every new free tool before anyone else.
If this is the call in front of you, let's talk it through.
A focused 30-minute conversation. No pitch. Just a direct discussion about the decision and whether I can help.
Book a call with Danny →
Danny Feldman is a 3× founder and the advisor behind Feldman Advisory. He founded and scaled Stellar to $100M in annual revenue across 15 states, raised $55M+ as a founder, and used AI to double revenue before exiting to private equity. Earlier, he recruited and trained 55,000+ sales reps at Cutco and grew a division from 1 to 58 locations. Operator to operator, he now helps founders of $1M–$20M companies get out of the middle of the business, build the leadership team, and raise capital. Book a 30-minute call →