The New 90/10 Rule for AI: What to Build vs Buy
The old rules don't apply anymore. SaaStr's updated framework for AI investment changes the fundamental question companies should ask about technology.
The new 90/10 rule: buy 90% of your AI capabilities off-the-shelf, build only the 10% that delivers outsized, proprietary value.
Why Buy Most of It
The economics have shifted dramatically:
Agent tooling makes complex builds feasible. What used to require months of development now takes weeks. Tools like Claude Cowork enable rapid implementation of sophisticated capabilities.
Maintenance costs add up. Custom apps aren't just build costs—they're ongoing commitments. Security updates, bug fixes, feature additions. Every line of code you write is code you maintain.
Speed matters. Buying means deploying today. Building means months of development before value materializes.
What to Build
The 10% worth building creates:
Proprietary differentiation. Capabilities competitors can't easily replicate. This is your defensible ground.
Core business logic. What makes your business unique—the workflows and processes that define your competitive advantage.
Deep integration. Systems that connect your critical data and processes in ways generic tools can't match.
The Vibe Coding Question
Rapid, low-code development has emerged as a legitimate approach. "Vibe coding"—building quickly based on intuition rather than detailed specifications—works for some use cases.
But it has limits. Security considerations multiply with custom builds. The time sink of maintaining custom apps can outweigh initial speed advantages.
The Stakes
Here's the reality: AI is now table stakes.
Products without meaningful AI risk losing customers. Competitors with AI capabilities will outperform those without.
But not every company needs to build AI from scratch. The winners will be those who combine excellent AI integration with strong domain expertise—not those who spend resources reinventing commoditized capabilities.
Applying the Framework
When evaluating any AI investment:
- Could we buy this capability?
- If we build it, what's the ongoing maintenance cost?
- Does this create proprietary value competitors can't match?
- Is this core to our business or peripheral?
The answer to these questions should guide whether you build or buy.
Stay ahead of AI trends. tldl summarizes podcasts from builders and investors in the AI space.