AI Voice Recorder for Transport Planners: Consultation Evidence, Model Assumptions and Decisions

Transport-planning discussions combine evidence types with different strengths and limits. A resident’s experience, traffic count, collision record, policy objective and model forecast may all matter, but they are not interchangeable.

Define the planning question

State whether the session informs an option appraisal, access review, consultation response, modelling assumption, safety assessment or formal decision. A defined question prevents a broad transcript from becoming an unfocused evidence store.

Use a claim-source-boundary structure

For each material point, record:

• claim, concern or proposed outcome;
• speaker or source;
• affected user group or transport mode;
• location and time period;
• whether the evidence is reported, observed, surveyed, modelled or policy-based;
• assumption and source date;
• uncertainty or limitation;
• analysis or decision required.

Protect consultation meaning

Do not use frequency as the only measure of importance. Accessibility, safety, severance and unequal impacts can be material even when raised by fewer participants.

Preserve who is affected, under which conditions and why the concern matters. Do not turn one participant’s experience into a statistically representative conclusion.

Keep model outputs in context

For modelling discussions, record the scenario, input set, base year, forecast period, geographic boundary, sensitivity and known limitations. Model output is conditional evidence, not an observed fact.

When an assumption changes, link the change to affected options and conclusions.

Separate four layers

• Evidence: consultation, survey, count, policy, model or site observation.
• Interpretation: what the planning team believes the evidence means.
• Option: an intervention under consideration.
• Decision: the approved direction, authority and date.

AI summaries can collapse these categories and present a discussed option as agreed policy.

Record trade-offs and distributional effects

A scheme may improve journey time for one group while reducing access, safety or amenity for another. Record who benefits, who may be adversely affected, the supporting evidence and what mitigation or further testing is required.

Convert evidence gaps into actions

For every material gap, identify the analysis required, owner, decision route, response date and communication method. Replace vague actions such as “consider feedback” with a specific output.

Use a consultation-to-decision workflow

1. Confirm authorised recording and privacy process.
2. State the planning question and evidence scope.
3. Capture source, place, time and affected group.
4. Label evidence type.
5. Correct names, road references, figures and speaker attribution.
6. Compare discussion with current documents and datasets.
7. Record uncertainty, trade-offs and minority impacts.
8. Update consultation, assumption or decision records.
9. Assign analysis and communication actions.
10. Delete temporary audio under the retention plan.

Frequently asked questions

Can a transcript replace the consultation record?

No. It can support the record, but contributions, analysis, responses and decisions still require the approved process.

Should frequently repeated comments receive automatic priority?

No. Frequency is context; severity, accessibility, affected groups and evidence quality also matter.

Can AI summarise model results?

It can draft a summary, but scenario, inputs, dates, boundaries and limitations must be checked.

Can AI select the preferred option?

No. Professional analysis and formal governance determine the decision.

Useful resources

• GOV.UK Transport Analysis Guidance
• AI Voice Recorder for Urban Designers
• AI Voice Recorder for Community Engagement Officers
• Assess Halo against the transport planners workflow matrix

Planning-record checklist

• Planning question and boundaries clear
• Lived experience, surveys and model evidence separated
• Assumptions dated and sourced
• Accessibility and minority impacts preserved
• Trade-offs visible
• Options separated from decisions
• Evidence gaps have owners and routes
• Formal records updated