Adaptive prompting
Timing, wording, frequency and format can respond to the task, execution history and recurring friction.
Technology and AI
Technology should carry support, not replace judgement.
Personalisation is becoming possible at a cost and scale that were previously impractical. That changes what organisations can design; it does not remove the need for evidence, judgement or governance.
Why the discipline is becoming more feasible now
Advances in AI, automation, connected systems and adaptive interfaces are reducing the cost and complexity of delivering person-specific execution support. This does not eliminate the need for human judgement, evidence or governance. It makes it increasingly practical to adapt selected components of the execution pathway at a scale that was previously difficult to achieve.
Historically, organisations often standardised not only the required outcome but nearly every part of the path used to produce it. Individualised workflows, prompts, feedback, interfaces and decision support were expensive to design, administer and maintain.
Standardisation remains essential wherever consistency protects safety, legality, ethics, evidence quality or minimum performance standards. What has changed is the growing ability to personalise selected parts of the execution pathway without changing the required outcome.
Enabling capabilities
What technology may change in the execution pathway.
Each capability is a design possibility to test, not evidence that personalisation will necessarily improve an outcome.
Dynamic sequencing
Complex work can be broken into different sequences while preserving quality and compliance requirements.
Personalised work presentation
The same information can appear as a checklist, summary, dashboard, decision tree, conversation or detailed guide.
Contextual decision support
Relevant information, constraints, risks and prior decisions can be surfaced at the point of choice.
Friction detection
Systems may identify recurring delayed initiation, rework, incomplete handoffs, missed follow-up or decision bottlenecks.
Adaptive scaffolding
Support can increase when difficulty occurs and reduce as capability and reliability improve.
Continuous feedback
Execution data can test whether an intervention changes behaviour and outcomes rather than relying only on self-report.
Coordinated support
Technology can help coordinate support from the individual, managers, teams, processes, training and specialist professionals.
Long-term model
An evidence-informed adaptive system.
A proposed feedback loop that combines context, observed execution and outcome evidence without treating any single input as a diagnosis.
- 01
Context
Person profile, role outcomes, task demands and known friction.
- 02
Observation
Execution behaviour, intervention history and outcome evidence.
- 03
Adaptive support
Proportionate support selected for the current task and context.
- 04
Measurement
Behaviour and outcomes observed against fixed requirements.
- 05
Refinement
Assumptions and support revised as evidence accumulates.
Safeguards
Personalisation without surveillance or unchallengeable judgement.
The risks are part of the research problem, not an afterthought to implementation.
No employee surveillance
Collect only data proportionate to a stated purpose, with meaningful transparency, consent and challenge pathways.
No automated diagnosis
Observed behaviour and self-report must not be converted into unsupported psychological or clinical judgements.
No management substitute
Technology should augment—not erase—appropriate human judgement, support and accountability.
No extraction logic
Adaptation should not become a mechanism for extracting more labour without regard for wellbeing or fair work design.
No unchallengeable decision
People should be able to inspect, correct and override consequential interpretations of their behaviour.
No unsupported high-stakes use
Employment decisions require sufficient evidence, transparent governance and appropriate human review.
Adaptation should support reliable performance and capability development, not create dependency or remove appropriate human accountability.
The Institute’s research agenda includes technology-specific questions about prompt dependency, automation bias, autonomy, surveillance and contestability.