Build versus buy

Our position. We hold that AI collapsing the cost of building shifts the build-or-buy equation (opens in a new tab). Bought packaged solutions give temporary speed, and built internal capability gives a permanent advantage. Most of the agentic tooling a team buys will not outlive the next model generation (opens in a new tab).

The Build or Buy Framework turns that into a test for each case: build what differentiates the business and what your team can understand and operate. Business software full of rules nobody on the team knows fails the second half of that test.

Does buying software still make sense when anyone can build?

The Silicon Valley Product Group analysis of 19 August 2025 qualifies our position for business software (SVPG - Build vs Buy in the Age of AI (opens in a new tab)). It places English-language builders such as Lovable and Bolt in a lineage that runs from VisiCalc in 1979 through Visual Basic and the low-code wave. What is new is that people without a technical background can now build. The common inference, that everyone will build, nobody will buy and SaaS is doomed, it calls almost certainly wrong. Procurement, invoicing, payroll and customer relationship management encode thousands of rules on policy, compliance, security, law, finance and pricing. Most non-technical builders do not know these rules exist. Engineers struggle too, because the rules are buried in code and the people who defined them have often moved on. That embedded knowledge is the same reason technical debt is hard to unwind, and the reason strong SaaS vendors are not about to disappear.

What changes is how bought components are consumed. The article reads MCP, which Anthropic introduced on 25 November 2024 (Anthropic - Introducing the Model Context Protocol (opens in a new tab)), as the missing protocol that lets software, and not only people, read and control business services. Build versus buy then becomes “yes to both”. Companies keep buying complex component services, and those components are designed to be driven by humans and by AI agents alike, whether the vendor, a systems integrator or the customer builds the agent. The lasting lesson from user programming is that the hard part is rarely building the solution. It is discovering the right solution to build (SVPG - Build vs Buy in the Age of AI (opens in a new tab)).

Related

• Build or Buy Framework: the four dimensions and five cost levels that decide each case.
• Architecture work in the agentic era: why design judgment stays scarce when building gets cheap.
• The software-factory pattern: the capability a team builds for itself instead of buying it.
• Pricing agentic coding: what the bought half of the agentic tooling costs.
• Protocol and context cost: what driving bought services over MCP costs an agent in context.

Resources

Articles

• Build vs Buy in the Age of AI (opens in a new tab) - Silicon Valley Product Group article arguing that business software encodes rules builders never see, so buying continues with agent-driven components.
• Introducing the Model Context Protocol (opens in a new tab) - Anthropic’s announcement of the Model Context Protocol, the open standard that lets software, not only people, read and control services.

Posts

• Shifts the build-or-buy equation (opens in a new tab) - LinkedIn post arguing that cheap building trades the temporary speed of bought solutions for the permanent advantage of internal capability.
• Will not outlive the next model generation (opens in a new tab) - LinkedIn post arguing that most agentic tooling a team buys today will be obsolete after the next model generation.