The AI install pack goes into the AI your team already uses, then paste in your OKRs for a review, or ask it to help write new ones. It coaches against a real methodology instead of the average advice the internet trained it on.
OKRs get one big thing right: they separate the ambition (the Objective) from the evidence it is working (the Key Results), and they pull an organisation toward outcomes instead of a list of tasks. Done well, they create focus and a shared language for progress.
Ask a generic LLM to write OKRs and it will hand you tidy task lists with round 70% targets, the two habits most correlated with OKRs failing. It has read a million blog posts and gives you their average, and the average is exactly what fails. The upgrade replaces that with the reasoning a good coach uses.
A task completes whether or not it worked. A behaviour change tells you the strategy is true. "Launch onboarding v2" is a task; "new-user activation from 46% to 60%" is the outcome it was supposed to move. Ask of every Key Result: what would a customer or the business do differently if this succeeds?
Most teams discover the real blocker at the post-mortem. Name it up front instead. The research on mental contrasting is unusually clear: pairing the goal with the honest obstacle is what converts ambition into commitment, because the plan now has to survive contact with what is actually in the way.
A percentage tied to reward teaches teams to sandbag: they write targets they expect to hit and call it ambition. Grade the quarter on judgement instead, with the evidence on the table. A well-reasoned miss can grade excellent; a sandbagged hit can grade poor. Honesty stops costing anything.
Focus is the entire point of the framework, so three objectives with nine key results is a to-do list wearing an OKR costume. One objective a team can recite, with two or three key results, beats a portfolio nobody can hold in their head.
The before is a task with a date; you can finish it and learn nothing. The after names the behaviour the work is a bet on, and the obstacle most likely to stop it, so the quarter teaches you whether the strategy was right.
Same files, three tools. Pick yours. It really is about two minutes.
Before you install, ask your AI to review one real goal and screenshot the answer. Install, ask again, screenshot the difference. That gap is the point. Try: Paste one Key Result and ask: is this a task or an outcome, and if it is a task, what customer or business behaviour is it really trying to move?
The AI install pack comes with From Trading to Scaling: an hour to read, five working templates, the instructions and knowledge files for Claude, ChatGPT or Copilot, and a conversation with Matt about any of it.
Free things get downloaded, not installed. A hundred pounds is the smallest number that makes somebody open the file, and this is the only place the pack is available.
The pack carries the principles and the working templates: the book, the strategy one-pager, the OKR template, the idea sheet and the bonus calculator, plus the knowledge files that make your AI coach against a method rather than the internet average. That is yours to keep, and it is enough to install all six things without us.
What it is not is the whole machinery. The playbooks, scoring scales and the 140-plus linked concepts, installed into your own tools and kept current, are the licensed layer above this. And the human work, the training, facilitation, coaching and mentoring that gets a company past its first messy quarter, is our people. Knowledge does not install courage, or referee a hard conversation. We would rather say that here than let you find out.
// one install, or the whole companyOne install helps one person. The advantage is your whole company running on a shared operating system: the same method in the AI everyone already uses, at every level, with routines and automations carrying the repeatable work, and your people freed to find and compound the knowledge and IP only you have.
One install at a time does not create that. Building the operating system does, and that is what ZOKRI does. See the method we install and the platform it runs on.
Most rollouts fail in a handful of named ways: task-based Key Results, round scoring tied to bonuses that teaches sandbagging, too many goals, and no obstacle named up front. Each has a fix, and the install pack flags them on sight. The deeper failure is treating OKRs as a reporting format rather than a reasoning system.
Almost never at team level. Cash on a team’s own goals kills the candour and ambition the whole system runs on. Grade the quarter on reasoning, and keep the few company-level strategic grades, read by named humans, as the only ones near money.
Fewer than you think. One clear objective per team, with two or three key results, is usually right. If nobody can recite the portfolio from memory, it is a to-do list, not a set of priorities.