Start with 5. Not 40. The companies that get real results from automation pick the handful of roles that carry the most volume and pain, get those working, and ignore everything else until they do. The ones that try to automate every department at once end up with a pile of half-built pilots and a team that's stopped believing in any of them.
The problem: the menu is too long
There's never been a longer list of things AI can do for a business. It can schedule, confirm, draft, summarize, reconcile, estimate, follow up, qualify and report. Across six departments, you can easily count 40 roles AI could take on.
That's exactly the problem.
When everything is possible, founders try to do everything. I see the same pattern over and over. The CEO comes back from a conference fired up. Every department head gets asked for an AI idea. A few vendors get signed. Pilots start in sales, marketing, ops, finance and customer service all at once.
Six months later, nothing is finished. Each pilot is owned by someone who already had a full-time job. The team quietly goes back to the old way because the old way works.
If everything is important, nothing is important. Automation is no exception.
Why going wide fails
Three reasons.
Attention is the real constraint. Every automation needs someone to define the process, test it, fix what breaks and train the team. Your best operators have maybe one of those projects in them per quarter on top of their day job.
Broken processes get automated faster, not fixed. If nobody has written down how a job gets estimated today, automating it just makes the mess run quicker.
Trust is built one win at a time. Your team will adopt the second automation because the first one made their week better. If the first five are all half-working, you've burned that trust before you've earned it.
What focus looks like in practice
Picture two companies of the same size, with the same budget and the same list of 40 possibilities.
The first one goes wide. Every department gets a pilot. Twelve automations start in the same quarter, each owned by someone with a full-time job already. By month six, four are half-working, eight are stalled, and the team has learned that "AI project" means extra work.
The second one picks four places where the volume is highest and people spend the most hours on work that doesn't need their judgment. Scheduling. Estimating. Status updates to customers. Invoice follow-up. One owner each. One number each.
By month six, the second company has four things running every day and a team asking what's next. The first company has a slide.
The difference isn't the tools. It's the discipline to say no to 35 good ideas so 5 of them actually get finished.
How to pick your five
Here's the filter I use with founders.
1. Find the $5-an-hour work your $500-an-hour people are doing. Walk through each department and look for the tasks that are high in volume and low in judgment. Scheduling, data entry, status updates, first drafts, chasing documents, reconciling invoices. Pay special attention to anything your most expensive people are still doing by hand.
2. Run the 90-day test. For each candidate, ask: "If this ran on its own tomorrow, would we feel it in revenue or capacity inside 90 days?" Anything that gets a "maybe" goes to the bottom of the list.
3. Check the process is real. Can someone write down, step by step, how the work gets done today? If not, fix the process before you automate it.
4. Give each one an owner. One name per automation. Not a committee.
You'll usually end up with 5 or fewer that pass all four. Good. That's your plan.
What to do this week
Crawl before we walk. Pick the single strongest candidate from your five and put it on the agenda for your next leadership meeting. Name the owner, the start date and the one number that will tell you it's working, whether that's hours saved per week, jobs handled per person or days cut from your close.
Say it this way to your team: "We're doing one of these, and we're doing it right. When it works, we'll pick the next one."
Hold the rest. They'll still be there next quarter, and your team will be a lot more willing to take them on after the first one lands.
The Automation Blueprint
The Automation Blueprint lays out 40 roles AI can take on, organized across 6 departments. Scroll the whole list, then select only the ones that pass the 90-day test. The tool builds a prototype of what your company looks like with those roles running, so your leadership team can see the change in one view. If you find yourself selecting more than 5, go back and cut. The discipline is the point.
Open the Automation Blueprint →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 →