I Don't Know What to Build with AI (And That's the Right Place to Start)

“I don’t know what to build with AI.”

That’s the most common thing I hear from people who are curious about AI tools but haven’t shipped anything yet. They’ve watched the demos. They’ve played with ChatGPT. They understand, in theory, that this technology can do remarkable things. And then they sit down to do something with it and draw a blank.

Here’s what I want you to know: this is not a creativity problem. It’s not a motivation problem. It’s not even an experience problem. It’s a sequencing problem. Most people try to answer the wrong question first, and then wonder why they’re stuck.

This article explains what the right question is, why the blank-page feeling is actually a signal worth paying attention to, and how to get from “I don’t know what to build” to a specific, worthwhile first project.

Why “I don’t know what to build” is the right place to start

The people who jump straight from “I want to build something with AI” to opening a tool and starting to prompt are usually the ones who end up with a half-finished side project and nothing to show for it three weeks later.

Not knowing what to build forces you to slow down and think clearly about the problem before touching the tools. That upstream clarity is what separates projects that ship from projects that stall.

There’s a real cost to skipping this step. If you pick the wrong problem, you can build a technically functional thing that nobody, including yourself, actually uses. You’ve spent time and energy, and the output teaches you nothing except that the wrong problem produces the wrong tool.

Starting from “I don’t know” is an honest acknowledgment that you haven’t done the thinking yet. That’s not a failure state. It’s the beginning of the process.

The reason most people stay stuck

When people say they don’t know what to build, they usually mean one of two things.

The first is that they’re starting with the technology instead of the problem. They’re thinking, “What could I do with Claude? What could I automate? What’s a cool use of AI?” These are seductive questions, but they rarely lead anywhere useful. Technology-first thinking produces demos and experiments. Problem-first thinking produces tools.

The second is that they’re setting the bar too high. They’re imagining their first build should be impressive, novel, or scalable. Something worth posting on LinkedIn. Something a friend would tell them is clever.

Your first build doesn’t need to be any of those things. It needs to solve one real problem that you personally have. That’s it.

The question that actually works

Instead of asking “what could I build with AI?”, ask: what would this replace?

Specifically: what are you currently doing manually, repeatedly, that a tool could do instead?

That question changes everything. It shifts your attention from hypothetical possibilities to actual friction you’re already experiencing. The problem already exists. The tool you’re imagining would eliminate something that currently costs you time or effort.

This is the question builders ask. Not “what’s an interesting application of AI?” but “what am I doing by hand that I shouldn’t have to do by hand?”

Write down your honest answer. If you sit with that question for ten minutes, you almost certainly have more than one answer. Most people do.

What non-developers are building

To make this concrete, here are four categories of things that people without traditional development backgrounds build well with AI tools.

Internal tools. Something you or your team uses to get work done. A dashboard that pulls from multiple data sources. A form that routes requests automatically. A generator that produces first drafts of a recurring deliverable. These tools save hours each week and the value is visible immediately.

Automations. A workflow that currently requires a human to initiate it. If you’ve ever thought “I have to remember to do X every time Y happens,” that’s a candidate for automation. Summarizing documents. Formatting data. Sending follow-ups. Moving information from one system to another. Automations free up attention.

Content systems. A structured, repeatable process for producing or repurposing content. Not a one-time prompt, but a system: something that takes an input (a transcript, a brief, raw notes) and reliably produces an output (a post, a summary, a draft) in a consistent format and voice.

Client-facing tools. Something your customers or prospects interact with directly. A calculator. An assessment. A configurator that helps people understand their options. These tools provide value and signal expertise at the same time.

None of these require deep technical skill to build. They require knowing what problem you’re solving.

Why the blank-page feeling eventually goes away

Once you orient around problems instead of possibilities, the blank page disappears. You stop looking for an idea and start looking at your actual work.

Most people have more build candidates than they realize once they start asking the right question. The recurring thing they do every week that takes an hour. The report they generate manually every Friday. The answer they write from scratch every time someone sends a particular type of request.

These aren’t glamorous problems. They’re real ones. And real problems produce real tools.

The first project doesn’t need to change your business. It needs to work. A tool that saves you 45 minutes a week is a real result. A system that produces a consistent output you used to build by hand is a real result. These early wins build the judgment and confidence that make the next project easier.

What to do right now

Pick a ten-minute window. Write down every recurring task in your work that you do manually more than twice a week. Don’t filter. Don’t evaluate whether AI could help yet. Just list them.

Then go back through the list and ask, for each one: what would it mean if a tool did this instead of me? What would the tool take as input? What would it produce?

You’ll find at least one candidate. Probably more.

That’s your starting point. Not a technology. Not a tool. A specific problem, a specific output, and a clear understanding of what it would replace.

Start with the thinking

The blank page doesn’t mean you have nothing to build. It means you haven’t asked the right question yet.

Spend ten minutes on that question: what are you doing manually, repeatedly, that a tool could do instead? The answer is already somewhere in your work. You just have to look at it the right way.

If you want a structured way to work through this, the 5-day AI Build Challenge is designed exactly for this moment: five short emails, one idea per day, no tools required. By Day 5 you’ll have a specific problem and a two-sentence project brief. That’s enough to start.