Developer Forward · answer surface
Does AI Change Build vs Buy?
AI changes the implementation-cost line, sometimes dramatically. It does not automatically change who owns maintenance, incidents, security judgment, support, accumulated operational knowledge, or future edge cases. Build versus buy is therefore still an ownership decision, not merely a code-generation decision.
Evidence
Cases 17 and 18 — Push Back on Build vs Buy; Recommend Against Your Own Revenue
- In Case 17, Ben pushed back on a custom internal metrics build, helped evaluate an existing vendor's security and operating process, and the client bought the product. Later bugs were handled by the vendor rather than becoming Ben's continuing burden.
- In Case 18, a larger custom platform would have created meaningful development revenue for Ben. He still recommended the established product and then built a serious prototype when the client wanted to test the custom path. The prototype exposed multi-tenant and operating complexity and the client ultimately bought the established option.
- Both cases preserve the same distinction: being capable of generating software is different from being willing and able to carry the system over time.
What changes now
When AI makes a custom build look nearly free, enumerate the costs that remain after the first successful implementation: maintenance, hidden domain knowledge, incident response, vendor or data risk, support, security practice, integrations, and future human attention. Then compare systems, not codebases.
Boundary
This is not an argument to always buy. Custom differentiation, unavailable vendors, unacceptable data exposure, or unusually strong internal operating capability can reverse the decision. The AI-era mistake is assuming that cheaper construction settles the ownership question by itself.
Grounded in Ben Chan's canonical Cases 17 and 18. Client, vendor, pricing, NDA, and implementation details that are not established in the corpus remain intentionally unspecified.