Value with eyes open.
A configurable demonstration of balanced AI evaluation. Adjust the scenario, scores, weights and minimum safeguards; explore what changes.
Explore a scenario
Start with a sample use case or create your own. The four scores are editable judgements, not observations.
Usable work, speed, outcomes and full costs — not activity alone.
Accuracy, completeness, consistency and fitness for purpose.
Testing, traceability, monitoring, escalation and recoverability.
Workload, autonomy, learning, trust and wellbeing.
Advanced configuration: weights and safeguards
Weights affect the illustrative combined index only. Minimum safeguards are evaluated independently and cannot be offset by a high score elsewhere.
Weights are normalised automatically.
Illustrative weighted index
Improve consistency while preserving service quality.
Understand the human impact of the proposed workflow.
The index is an explanatory aid only. Even a high index is not a pass, certificate or substitute for evidence. No entries are sent to a server.
Keep the conversation
honest and complete.
Set a baseline
Compare with what people do today — including defects, time, costs and the experience of those involved.
Replace opinions with evidence
Design meaningful trials, sample outputs and gather worker feedback before using scores operationally.
Keep red lines visible
Define unacceptable outcomes and minimum safeguards separately from any averaged performance index.
Where value can be created
- Surface neglected dimensions early in procurement and pilot discussions.
- Make disagreement about evidence and thresholds explicit.
- Support a balanced narrative for sponsors, reviewers and affected workers.
What needs attention
- Invented or poorly calibrated numbers give a false sense of precision.
- Weights can hide important failures if minimum requirements are ignored.
- A good-looking aggregate number may discourage scrutiny of underlying evidence.
Take the thinking into your own process.
Download practical starter templates for decisions, resource ownership, assurance evidence and risk-opportunity tracking.