Your OKRs, stress-tested by AI.

This kit is a working slice of ZOKRI's Operating System for OKRs — the public layer: core definitions, our ten guiding principles, and a reviewer that applies them to your drafts without flattery.

It is genuinely useful. It is also nowhere near the full version. The complete system is 50+ linked concepts covering the things that determine whether OKRs actually work: how to run check-ins that change confidence rather than report status, how to form and resource OKR teams, how to grade a quarter in retrospectives, how to build metric trees and find the levers, how discovery runs alongside delivery (spoiler: not inside your OKRs), plus the templates, playbooks and scoring scales that make all of it repeatable. That system is licensed and installed into your organisation's AI tools, so every team gets these answers, every day, in context, and it stays current as the methodology evolves.

And a tool that critiques goals is not the same thing as an organisation that can set and achieve them. That capability is built by humans, which is what our training (skills, taught properly), coaching (capability embedded alongside your teams, quarter by quarter) and mentoring (senior support for the leaders carrying the goals) exist for. The Coach makes your drafts better. The people work makes your company better at drafting. They are complementary, not alternatives.

If you never buy anything from us, this kit will still improve your OKRs. That's the deal.

Paste the whole draft, plus context. The Coach is only as sharp as what you give it:

• The Objective and every Key Result, exactly as drafted — don't tidy them first.
• Baselines and how each metric is defined ("activation" means nothing until you say what counts).
• Who owns it, what else the team is carrying, and the time period.
• What evidence sits behind each target, if any. "None yet" is a fine answer; the Coach will work with it.

The Coach will ask questions before judging when information is missing. Answer them — the review after your answers is the real review. Expect two or three rounds before a "Ready".

Each OKR gets a verdict, what works, issues ordered by severity, a rewritten version, and the questions to resolve before launch. The overall grade is one of:

• Ready — launch it, resolve the listed questions in week one.
• Needs work — the outcome focus is right; the measures or evidence aren't yet.
• Rethink — it's a work plan or a discovery project wearing an OKR costume. Rebuild around what would be different if you succeed.

A "Rethink" is not failure. Most first drafts we see professionally would grade Rethink. That's precisely why reviewing before the quarter starts is worth doing.

• Draft from zero: give it your strategy context and the change you want; ask for three candidate OKRs to react to.
• Task-to-outcome conversion: paste your roadmap list and ask what would be different if each item succeeded.
• The focus test: paste everything your team plans this quarter and ask which single OKR is the wildly important one.
• Evidence grading: ask it to grade the evidence behind each target and tell you what would raise the level.
• Pre-mortem: "It's the end of the quarter and this OKR failed — what most likely killed it?"

The Coach reviews artifacts. It cannot run your check-in, resource your team, resolve the conflict between BAU and the OKR, coach a first-time OKR lead through a bad week, or make your leadership team actually commit to fewer goals. Ask it about those topics and it will tell you plainly that they're covered by the full system and the humans who install it.

That's not a limitation we forgot to fix. Tools review; people build capability.

• Too agreeable? Say: "Be tougher. Assume I'll launch this tomorrow unless you stop me." Or upgrade the model.
• Inventing baselines or metrics? It's been told not to guess — remind it: "Ask, don't assume."
• Generic OKR advice creeping in? Check the knowledge files are actually loaded; the instructions tell it our files win over general OKR lore.
• Reviews drifting over long chats? Start a fresh chat per OKR set. The project/GPT keeps the setup.