Kaivant-O: The Organisation Instrument
A non-compensatory composite built around two structural axes: Leverage Architecture and Organisational Capital. Neither can compensate for the other. An organisation cannot average its way to a high score.
The Kaivant Score is structured so that high performance on one axis cannot compensate for low performance on the other. The composite score floors at the lower of the two axis scores, with the higher axis able to contribute only a limited premium above a defined threshold.
This is not an arbitrary design choice. It is the mathematical expression of the framework's central claim: an organisation with exceptional Leverage Architecture and depleted Organisational Capital is not well-positioned. It is generating strong current output while accumulating the structural fragility that will limit its capacity to sustain or adapt it. The score reflects that reality directly.
What the organisation's design has produced.
Leverage Architecture measures the structural conditions that convert AI investment into compounding leverage. High LA indicates that AI integration has genuinely reshaped how the organisation coordinates, decides, and operates, not merely how it executes individual tasks.
Coordination Efficiency
The degree to which AI integration has reduced friction in information flow, decision-making, and cross-functional coordination. High CE indicates that the information-transfer infrastructure built to manage information scarcity has been materially reduced because AI makes organisational data continuously visible and interpretable.
On a Monday morning
The symptomThe calendar is full and the work is not moving. More of the day goes to keeping everyone aligned than to producing anything.
On a real teamEvery decision needs three sign-offs and four status meetings before anything ships. A new joiner spends their first month learning who to ask rather than how to do the work.
The smallest first moveTake one recurring decision that needs a sign-off, pick a low-risk category, and remove a layer. Name the person who can now decide alone and watch what happens to cycle time over a fortnight.
Decision Velocity
The speed and level at which decisions are made, and how much AI has shifted the threshold for autonomous versus escalated decisions. High DV achieved through information quality and decision clarity is the target. High DV achieved by shortcutting analytical engagement is the failure mode.
On a Monday morning
The symptomDecisions sit. Not because they are hard, but because nobody is sure who owns them, so they drift upward until someone senior clears the queue.
On a real teamThe same kind of decision takes two days one week and three weeks the next, depending on who is in the room. People escalate to be safe, and the escalation point becomes the bottleneck.
The smallest first moveFor one recurring decision type, write down in a sentence who decides, what they can decide without asking, and when they must consult. The aim is removing the wait for permission on the decisions that were always going to be approved.
Autonomous Workflow
The proportion of workflow that operates without human orchestration at handoff points. AI-enabled automation reduces handoffs not primarily by executing tasks but by maintaining context across tasks, removing the need for human re-orientation at each transition.
On a Monday morning
The symptomAI is in the building, but it touches isolated steps. Every handoff still needs a person to copy, paste, reformat, and pass along. A tool saved a task and the workflow did not get any shorter.
On a real teamA seven-step process has three automated steps and still moves at the speed of the four manual handoffs between them. Every exception lands on the one experienced person who understands the whole flow.
The smallest first moveMap one end-to-end workflow and mark every point where a person picks the work up only to pass it along unchanged. Design out the handoff that waits the longest first, with the person who catches the exceptions in the room.
Leverage Trajectory
Whether leverage is building, plateauing, or declining: the directional signal. LT is the financial fingerprint of the compounding dynamic: reduced coordination cost enabling more redesign iterations per period, each creating the conditions for faster subsequent cycles. The lag/lead gap on LT is the most important single signal the instrument produces.
On a Monday morning
The symptomOutput per person is flat, even though the tools are better than last year. The gains showed up once, early, and then stopped. Nobody can say where the next one is coming from.
On a real teamThere was a real jump when AI first arrived, then the work settled back into its old shape. One function pulled ahead and the rest never caught up, so the average looks unchanged.
The smallest first moveFind the function that improved most and the one that improved least, and ask why the gap exists. Copy one specific working practice from the strong function into the weak one and measure the difference over a quarter.
What the organisation is building toward.
Organisational Capital measures the accumulated human capacity that determines whether Leverage Architecture gains are durable or brittle. Without a high and developing OC, high LA produces gains that are real but structurally fragile: optimised for today, with diminishing capacity to adapt to tomorrow.
Learning Velocity
The rate at which the organisation converts experimentation into codified, reusable knowledge. LV measures both how often the organisation runs experiments and whether those experiments convert into reusable knowledge. Reduced coordination cost expands the experimentation budget; LV measures whether the organisation is spending it.
On a Monday morning
The symptomThe same lessons get learned over and over. A problem solved in one corner of the business reappears in another six months later, because what was learned never left the room it was learned in.
On a real teamThe knowledge lives in a few people's heads and walks out the door when they do. Experiments happen, but the results are not written down, so the next person repeats them from the start.
The smallest first movePick one thing the team recently worked out and spend twenty minutes turning it into something reusable: a checklist, a template, a saved prompt. Put it where the next person will find it without having to ask.
Human Judgment Utilisation
The proportion of human working time genuinely exercising judgment rather than approving AI outputs. The primary risk of extensive AI deployment is not that the organisation will make mistakes. It is that it will eliminate the human capacity to catch them. HJU measures whether that error-correction capacity exists and is being exercised.
On a Monday morning
The symptomPeople are spending their judgment on work that does not need it, and rubber-stamping the work that does. AI outputs get approved with a glance, while skilled people retype things a machine could have produced.
On a real teamExperienced staff do rule-bound tasks out of habit, and the AI-assisted decisions pass review without anyone really looking. Override rates sit near zero, which sounds like trust but usually means nobody is checking.
The smallest first moveFind one decision where an AI output is accepted without real review, and one task where a skilled person is doing work a tool could do. Swap them, and notice whether the review actually changes the output.
Capability Development Velocity
The rate at which human capability is growing across the organisation. Human capabilities depreciate toward irrelevance when the threshold of what constitutes valued work shifts beneath them. CDV measures whether the organisation's human capital stock is appreciating or depreciating, and whether capability development investment is reaching the people it is designed to serve.
On a Monday morning
The symptomThe tools are getting better faster than the people are. AI is absorbing tasks, and the capability that used to come from doing those tasks is quietly not being built anywhere else.
On a real teamJunior staff used to learn by doing the work AI now does, and nothing has replaced that ladder. People can feel their skills depreciating against the tools, so they stop volunteering for the stretch.
The smallest first moveTake one role where AI has absorbed the routine work and name the higher-judgment capability it should now grow into. Give one person three hours a week to build it deliberately, with feedback from someone who already has it.
Human Dignity and Agency
Whether the design of work preserves the conditions for motivated, meaningful human contribution. High Leverage Architecture built on degraded human dignity is not a stable equilibrium. The lead indicators of HDA, psychological safety trajectory in particular, are designed to surface deterioration before it reaches the performance layer.
On a Monday morning
The symptomThe people operating the AI systems have no say in how those systems work, and no safe way to flag when they are wrong. Concerns get swallowed. The quietness reads as acceptance, right up until it reads as attrition.
On a real teamAI-driven changes arrive as announcements, not conversations. People have learned that raising a problem with a system is not worth the cost, so errors get worked around in silence instead of fixed.
The smallest first moveOpen one channel for the people closest to an AI workflow to say what is not working, and respond to the first thing they raise in a way that costs them nothing to have raised. This is the floor of the framework.
Below a defined HDA threshold, the Kaivant Score is flagged regardless of composite performance. An organisation where employees cannot challenge AI outputs, have no psychological safety to escalate concerns, and have no meaningful agency over their work has removed the early-warning infrastructure that prevents AI system failures from becoming organisational crises. No leverage score is an adequate offset for that condition.
What a Kaivant-O result looks like.
The profile below is an illustrative example showing the structure of a Kaivant-O output: composite axis scores, nine dimension readings with lag and lead, and the gap between them that is the instrument's most actionable signal.
Strong demonstrated past adaptation · thinning experimentation pipeline · declining proactive redesign rate. The intervention point is the lead reading, not the lag. This is the AA signal that most directly warrants action.
The leverage this organisation has built is real. Coordination is working well, decision-making has accelerated, and a meaningful share of workflow runs without manual handoffs at every step. The composite score reflects genuine structural progress. The constraint is on the Organisational Capital side: the accumulated human capacity that makes these gains durable has not grown at the same pace as the leverage the organisation is now generating.
The leverage trajectory and adaptation architecture readings are not two separate concerns. They are the same story at different timescales. Adaptation architecture is the mechanism through which the organisation generates each new cycle of improvement: running experiments, learning from them, and redesigning processes to produce the next gain. When that mechanism weakens, leverage trajectory is the first place the effect shows up, because the next wave of performance depends on the previous adaptation cycle having run.
Both are showing the same thing: the improvement cycle is slowing. Fewer experiments are being run, and what is being learned is not being converted into reusable knowledge at the rate the organisation's past performance required. The organisation is drawing on previous adaptation cycles without actively replenishing the engine that generated them.
The one dimension moving in the right direction is the rate at which people are spending time on genuine thinking work rather than approving AI outputs. That is precisely the input a recovering adaptation engine needs. It connects directly to two dimensions that matter most for recovery: the pace at which human capabilities are growing across the organisation, and the rate at which new learning is being captured and codified. Both of those are currently declining on their forward indicators. The judgment is present; the infrastructure to build on it is not yet in place.
The organisation already has the human capital to address this. A structured capability investment that connects the judgment currently being exercised to deliberate skill-building and systematic learning capture would address the two weakening forward indicators at once. Those two dimensions feed directly into the adaptation engine. When that engine is running, leverage trajectory stabilises naturally, not as a target pursued directly, but as the consequence of an improvement cycle that is actively generating the next wave.
The bridge between axes.
Adaptation Architecture occupies a structural position that neither axis fully occupies independently. It measures whether the organisation has the structural capacity to identify opportunities for improvement and implement them at declining cost per cycle: the mechanism through which the two axes compound each other.
AA is weighted more heavily in the Organisational Capital composite because the capacity to adapt is itself a form of organisational capital. Its contribution to the Leverage Architecture composite reflects demonstrated historical adaptation performance.
The critical diagnostic signal is the lag/lead gap: an organisation with strong demonstrated adaptation in the past and a thin pipeline of experiments and declining proactive redesign rate today is at peak performance with a depleting adaptation engine. This is the AA signal that most directly warrants intervention.
Each cycle of adaptation is not merely additive. Reduced coordination cost enables more redesign iterations per period, and each iteration creates the conditions for faster subsequent cycles.
On a Monday morning
The symptomImproving anything is expensive. A change that should take a week takes a quarter, because every redesign starts from scratch and nobody remembers how the last one was done.
On a real teamThe team knows what to fix and dreads fixing it, because the last process change cost three months and a lot of goodwill. Improvements happen only after something breaks, never before it does.
The smallest first moveTake the last improvement you made and write down how you made it: what changed, who was involved, what it cost. Make that the template for the next one. The goal is for each redesign to cost a little less than the one before.
Not for ranking. For growth.
Each dimension score produces a diagnostic profile. Each profile maps to the intervention library at kaivantscore.com: concrete, evidence-grounded responses to each failure mode the instrument identifies.
The Kaivant Score is designed as a spot inspection, not a ranking system. It produces a lag reading (current architectural state) and a lead reading (trajectory). The gap between them is the most actionable output: an organisation with a high lag score and a declining lead score is performing well today and building the conditions for underperformance tomorrow.
Reassessment at six and twelve months tracks trajectory. The instrument is most useful longitudinally: not as a single score, but as a map of where the organisation is moving and why.
Where we are, honestly.
The Kaivant Score is currently at Stage 0: a theoretically grounded diagnostic with predictive validity under development. No outcome data has yet been collected. The permissible claim at this stage is: theoretically grounded diagnostic; predictive validity under development.
We publish our validation stages, evidence thresholds, and current standing openly. No stage is claimed without supporting data. A measurement standard that earns trust does not claim validity it has not yet demonstrated.
Stage 0: Theoretically grounded diagnostic. Predictive validity under development. No outcome data collected. Stage 1 threshold: n ≥ 50 organisations with 6-month follow-up.
Ready to understand your position?
The Kaivant Score launches on kaivantscore.com in 2026. The score is the agenda. The facilitated session is where a leadership team turns it into two or three agreed moves. Register your interest, or read what a session looks like.