A Prospect Sent Me Two Sentences. Three Hours Later, They Had a Working AI System.

I’m not a programmer. Keep that in mind as you read this, because it’s the whole point.

A few weeks ago a lead reached out by email. Two lines about their company. That was it. They sell services to companies that win large government contracts, and their entire business depends on finding out who just won what, before their competitors do.

Before I ever got on a call, I handed those two sentences to my Tableland Partners AI copilot and asked what it could tell me. It came back with their likely business model, an estimate of company size, a breakdown of their target audience, and a walkthrough of the public website where the contract data gets posted every day. It even flagged the security barriers it couldn’t get past, which told me something useful about why this problem existed in the first place. I walked into the discovery call able to talk intelligently about an industry I had never touched.

The problem they described

Their team was hand-searching through roughly 10,000 new files posted to a clunky government website every single day. Clicking through security banners. Scanning file names for clues. Downloading PDFs one at a time to check whether a contract was even relevant. Around 15 hours of skilled work every week, spent on something a machine should be doing.

You probably have a version of this in your business. Not government files, maybe. But some daily grind where a person scrolls through a big pile of stuff looking for the handful of items that matter.

What happened after the call

I got off the one-hour discovery call, opened a fresh conversation with my copilot, and typed the problem in plain language. Here’s the exact structure I used, and it’s the same structure I now teach every member of my AI biweekly Roundtable:

That last line matters more than anything else in the prompt. The AI wrote the code, then treated me like a smart person who had never seen a command line. Create this folder. Name it this. Paste this. Hit enter. At one point I named a folder wrong and got an error. I pasted the error back in and it just fixed it for me.

Three hours later, I had a fully built AI program doing everything their team had been doing by hand. The 15 hours of weekly work now runs in about 40 seconds.

Since then, I have built 4 such programs for customers across totally different industries. Turns out its a common problem: sorting through lots of information to get a specific task complete.

Then we put it in the cloud

One more question finished the job for several of my customers. I told it I didn’t want this running on my desktop. I wanted it running automatically every morning, sending an email to the customer with what they wanted. It laid out the whole path: sign up for a small cloud server, move the folder over, schedule the run.

Now nobody pulls a trigger. You just wake up to an email with everything in it.

The same pattern, simpler version

On a recent Roundtable call I demoed this exact pattern with a simpler example so members could watch it happen live with sharable data. Imagine a landscaping company. Their city publishes new building permits every day as a spreadsheet, and new construction is their best source of leads. Their rules from a discovery call might look like this: anything over 500,000 dollars is a hot lead. Anything from 100,000 to 500,000 is worth watching. Skip the rest. Flag the two contractors they already have relationships with at the top of every report. Estimate the landscaping budget at 3 percent of construction value so the report shows what each job is worth to them.

Same five-part prompt. Same result. A ranked call list every morning with dollar values attached, instead of an hour of scrolling.

What this actually means

The barrier to custom software used to be code. It isn’t anymore. The barrier now is whether you can describe your own business rules out loud. If you can explain to a new employee what makes a lead worth chasing, you can explain it to an AI. And the AI can build the thing.

That’s a strange sentence to write after 20 plus years in marketing. It’s also true, and the businesses that figure it out first are going to be very hard to catch.