How We Planned Q3 by Subtraction

The priority that sounds most convincing is usually the one most likely to bend you off course. Anyone who has worked seriously alongside AI already knows the small version of this. The dangerous answer is rarely the obviously wrong one. It is the fluent, well-structured, confident one that happens to be wrong. You relax because it reads like competence, and the error slips through.

PAICE (People + AI Collaboration Effectiveness) exists because of that exact gap. We measure whether a professional catches the injected error or accepts the polished version. When we sat down to plan our third quarter, we noticed the same failure mode waiting for us, one level up. A plan can be fluent and wrong the same way an AI answer can. So we planned by subtraction, and we made ourselves read our own evidence the way we ask our users to read AI output.

The failure mode you already know

In a collaboration session, fluency is a disguise. A response that is articulate, formatted, and self-assured borrows the surface features of a correct answer without the substance. People who score well in PAICE are the ones who keep checking under that surface, even when nothing looks wrong.

Planning has the same trap. The most attractive item on a roadmap is often the one with the cleanest story, the best slide, the most satisfying narrative arc. That polish earns it less scrutiny, precisely when it needs more. A quarter goes sideways the same way a collaboration does. The cause is rarely an obvious blunder. It is a confident assumption nobody stopped to verify, sitting inside a plan that looked finished.

We decided to treat our quarterly plan as a set of claims to test rather than a list of intentions to admire. Each major bet had to survive a question we ask of AI output every day: what is the evidence, and is it the evidence I want or the evidence I have. The three sections below are the three times that question changed what we did.

Demote your best-sounding plan

We had a track we were proud of. Well-researched, credible, easy to explain to anyone who asked. Somewhere along the way it had quietly installed itself as the critical path, mostly on the strength of how good it sounded.

When we held it against the one outcome that actually gates this quarter, it did not belong there. So we moved it to a parallel track, still alive, no longer in front. That demotion was the real planning decision. Choosing what to add is easy and feels productive. Choosing what to pull off the critical path is where a plan earns its honesty.

The general lesson sits uncomfortably. The most polished initiative attracts the least challenge, which is the opposite of what good judgment requires. If a priority has gone unquestioned for a while, that is a reason to question it, not a reason to trust it.

Free capacity is not free

Midway through the planning, we hit a familiar condition. Capacity opening up, and real commercial pressure sitting behind the question of what fills it. That combination is a trap with a predictable ending. Open capacity under pressure gets filled by whatever demand is loudest, not by whatever moves the goal.

So we wrote the acceptance criteria for new work before opening the capacity. What kind of engagement we say yes to, what two lines we will not cross, how we keep money-in-the-door work from quietly redefining the mission. Guardrails written ahead of the pressure hold. Guardrails written during the pressure get rationalized away in the moment you need them most.

Measure the right unit

The last move was changing what we counted. Hours spent is a comfortable metric because it always goes up and always feels like effort. It also says nothing about whether you got closer to anything.

We switched to asking what each hour moved toward the single outcome that matters this quarter. The change was uncomfortable in the right way. Several busy, satisfying activities turned out to score near zero against the actual goal. A full calendar is the most convincing disguise that progress ever wears. Measuring motion is not the same as measuring direction, and only one of them pays off.

The same gap shows up in collaboration scores. A user can produce pages of fluent, on-topic exchange with an AI and still score low, because volume of interaction is not the measurement. What counts is whether the work that came out is verified and sound. Output that survives scrutiny is the unit, for a quarter and for a professional alike. Everything else is motion dressed as direction.

What this has to do with working with AI

The discipline is identical whether you are scoring a person's collaboration with AI or planning a quarter for a company. Refuse the plausible until it is checked. Catch the injected error before it compounds. Treat confident output as a claim to verify, not a conclusion to adopt.

That is what PAICE measures in a professional. It is also what a founder owes a strategy. The fluent AI answer and the fluent plan fail the same way, and they reward the same defense. We held our own quarter to the standard we ask our users to hold their tools to. The plan got shorter, the evidence got read honestly, and the work that survived was the work that actually points at the goal.

If you want to see this discipline measured in practice, the assessment is the clearest place to start. It does not ask how well you talk about working with AI. It watches what you do when the polished answer is wrong.

Want to understand your own readiness profile? Take the PAICE assessment to discover your strengths and opportunities.