Build vs. Buy & Who to Hire
You can buy an agent platform, build on your own stack, or blend the two — and all three are right for different companies. The mistake is deciding by hype instead of by whether the agent's judgment and actions are core to your business.
Off-the-shelf agent platforms get you moving fast and are often right when the automation is a convenience on low-stakes, reversible actions. Building makes sense when the agent takes consequential actions in your systems, when reliability and safety are existential, or when scale makes cost per task strategic. A common middle path: buy the general framework and models, but build the parts that are yours — your tools, your gates, your evaluation, your reliability handling.
Whatever you choose, someone must own the agent's reliability and safety. That's the hire: not necessarily a machine-learning PhD, but an engineer who thinks in loops, tools, stopping rules, gates, and completion metrics — who treats "how do we know it finishes safely?" as their core job. The appendix gives you the questions to find that person.
Build or buy?
Two peers made opposite calls: one bought an agent platform, one built in-house, and both are happy. What single question decides which is right for you?
Ask: "Does the agent take consequential actions in our own systems, where reliability, safety, and cost are core to us?" If yes — it moves money, touches customer data, or its failures are existential — lean toward building (or buy-the-framework, build-the-tools-gates-and-evals), because you can't outsource ownership of actions taken in your name. If it's a convenience automating low-stakes, reversible work at modest scale, buy and move on. Either way, one rule holds: even if you buy, measure completion on your task suite and keep control of the gates — never let a black box take real actions unmeasured.
Two companies, two right answers
Two companies faced the choice. DeskZero, a small team, wanted an agent to tidy and route internal notes — low-stakes, reversible, not a differentiator. They bought an off-the-shelf platform, wired it up in days, and spent their engineering time on their actual product. Right call. MedLedger, a healthcare operations company, needed an agent taking consequential actions in patient-facing systems where a wrong or unsafe action was existential. They bought the general framework but built their own tools, gates, evaluation, and reliability layer in-house. Also the right call.
The difference wasn't budget or sophistication; it was whether the agent's actions were core and consequential. Both, tellingly, measured completion on their own task suite — DeskZero to keep the vendor honest, MedLedger to govern their own build and gates. The leadership lesson: decide build-vs-buy on how consequential the actions are, and measure the completion either way.
Tessera lands in the middle: it buys the agent framework and models but builds its own tools, gates, and evaluation — the parts tied to its systems and its risk. To own it, they hire one engineer whose whole job is "how do we know it finishes the job safely?" The next section is how they found her.
Quiz · Chapter 12
- Lean toward building when:
- A common sensible middle path is:
- Even if you buy, you should:
- The key hire is someone who: