The build-versus-buy decision has a simple rule that AI has not changed: buy unless you genuinely cannot. What AI has changed is the cost of the “build” side, which shifts some borderline cases.
When off-the-shelf is right
- A configured product covers the process well enough.
- The process is not a source of competitive advantage; it just needs to work.
- You want someone else to handle updates, security and support.
Most business processes are like this. Fight the urge to build.
When custom is right
- A process that is core to how you make money is genuinely specific to your business.
- Off-the-shelf tools force daily workarounds, spreadsheets alongside the tool, manual steps, data in the wrong shape.
- A new revenue line needs software that does not exist.
- Several disconnected tools cause errors that a purpose-built system would remove.
How AI changes the maths
AI reduces the cost of writing and testing code. That makes a focused custom build viable in cases where it previously was not, particularly internal tools and integrations. But understanding the problem, designing the system, connecting it and maintaining it are still most of the cost, and AI helps less with those.
The commitment
Custom software needs maintenance, can be over-built, and has a cost of ownership that continues after launch. Go in knowing that.
How XP Labs approaches it
XP Labs looks hard for a reason not to build, a tool that would fit, a process that could change, before recommending custom software. When it is the right call, the first build is deliberately small and focused on the highest-value part. The planning conversation tests the decision. See the custom software and MCP development overviews.