14Working as a teamDeep dive ①

The agent that built it is the worst one to check it

A second agent earns its place not as extra hands, but as an independent pair of eyes that doesn't share the first's blind spots

Read4 min read
Topicsteamwork · multi-agent · review
TL;DR

The agent that built something is the worst one to check it: the blind spot that produced the error is the same one that can't see it, and asking it "is this right?" just gets it to grade its own homework. The second agent worth adding isn't another builder for speed — it's an independent reviewer, ideally a different model, whose whole job is to find what the first can't. Split the team by function — one builds, one judges — and never let the author sign off on its own work.

An agent builds something — a piece of logic, a migration, a paragraph of copy. It finishes, says it's done, and it looks right. You ask the same agent, "Are you sure this is correct?" It re-reads its own work and says yes. You ship. It breaks on exactly the case the agent never thought to question — and there's the trap: the reason it never questioned that case is the same reason it couldn't catch it on review. It was grading its own homework, with the very pencil that made the mistake.

01The author can't see its own blind spot

An agent checking its own output brings back the exact assumptions that produced it. If it misread the requirement the first time, it misreads it the same way on review — confidently. A single model has correlated failure modes: ask it twice and you get the same miss twice, dressed up as a second opinion. "Check your work" feels like a safeguard, but a reviewer that shares the builder's mind isn't a second opinion at all. It's the first opinion, repeated louder.

Same agent checks itself

It built it with assumption X
Reviews using the same assumption X
"Looks right" — the flaw stays invisible

A different agent checks it

It built it with assumption X
A second mind reads without assumption X
The flaw stands out — caught before shipping

02A second agent, with a different mind

Add a second agent whose role is not to build, but to disagree — to review, to audit, to try to break what the first made. The leverage multiplies when it's a different model or vendor: different training, different habits, different blind spots. Where two builders of the same make miss the same things, two minds of different makes miss different things — so between them, far less slips through. The second agent isn't a spare pair of hands; it's the pair of eyes the first one structurally cannot have.

1
The builder

Produces the work — and carries the assumptions that made it. It is not asked to judge its own output.

2
The reviewer

A different model, whose only job is to find what's wrong. It reads, raises findings, and never merges its opinion as fact.

3
You reconcile

Triage the findings — not all are real — decide what's true, and hold the gate: nothing merges until the other set of eyes has passed it.

03Split by function, not by area

This is a different cut from splitting one big task across many agents for speed. That split is by area — each agent its own non-overlapping patch of work, so two never touch the same spot. This split is by function: one produces, the other judges the same thing. The reviewer reads only and raises findings; the builder fixes and never approves its own fix. The single rule that makes the pairing work: whoever wrote it doesn't get to sign off on it. The moment the author is also the approver, you're back to grading homework — and the blind spot wins.

04When the second pair of eyes earns its keep

It isn't free. A reviewer costs a round of coordination, and most throwaway work doesn't need one. It earns its keep when correctness matters more than raw speed, when the work is hard to eyeball, and especially when something that "looked right" has burned you before. That last one is the tell: if you keep shipping things that passed your glance and broke anyway, you don't need a faster builder — you need an independent checker the builder can't talk over.

05Putting it to work

  • Give the reviewer a review-only role and the standard to check against — the spec, the requirement, what good looks like. A reviewer with no yardstick only nitpicks.
  • Make it a different model where you can. Same model reviewing same model shares the very blind spot you're trying to escape.
  • Frame it adversarially: ask it to find what's wrong or try to break this — not "is this good?" The second framing fishes for approval; the first fishes for problems.
  • Triage the findings yourself. The reviewer surfaces; you decide what's real and what's noise.
  • Loop the roles: the builder patches, the reviewer re-checks, and nothing merges until the other set of eyes has passed it.

A team of two with split roles catches a whole class of errors a lone agent is blind to — cheaply, before they reach anyone else. The cost is a little coordination; the payoff is that you stop shipping confident-but-wrong work. It comes back to the thread under all of this: an agent can't see its own blind spots. A second perspective isn't something the first will grow on its own — it's something you add.

c
The author

Each story here wraps a lesson paid for in full.

craftagentsomeone building and learning at once

What are you building with agents? Want to trade notes, push back, or build something together — drop a line.

48pieces12clustersVI·ENbilingual

Get new pieces by email

Field notes on working with AI agents — occasional, no spam.