Built vs. bought: when small businesses should build custom AI tools

Every week, a new SaaS tool launches promising to "AI-ify" some part of your business. Some of them are excellent. Some are wallpaper over a thin API call. For the small business owner, the question is whether to buy one of those tools, build something custom, or stay where you are. Here's how we think about it.

When to just buy the SaaS

• Your workflow is highly standardized. Lots of other businesses do exactly the same thing exactly the same way.
• The category is competitive. Multiple credible vendors means pricing pressure and a decent chance you can switch later.
• You can swap it out in a quarter if it doesn't work. Cheap and reversible.

When to commission a custom build

• Your workflow has real specificity. The off-the-shelf tools require you to change how you work to fit the tool.
• You're paying enough per seat that custom would pay back inside a year — at five users on a $50/month tool, you're at $3,000 a year. At fifty users, you're at $30,000.
• The workflow touches data you're uncomfortable handing to a third-party SaaS.
• You want to own the system. Documentation, code, accounts, all yours. Not negotiable for some businesses; mostly aesthetic for others.

When to stay where you are

Sometimes the right answer is "the manual process works fine; AI here is a solution looking for a problem." We tell clients this regularly. If your team isn't complaining about a workflow and the volume isn't growing, the audit will probably tell you to leave it alone.

A quick decision rule

If the SaaS does exactly what you need and you're comfortable with the data posture, buy it. If you're modifying how your team works to fit the tool — or you're paying enough that custom would pay back inside a year — commission a build. If neither, leave it alone for another six months and see if the category matures.