Research suggests 50 to 60% of meaningful call detail is currently lost when forces rely on operator notes rather than full audio. West Yorkshire's Post Call Analysis pilot was built to address that gap — an SME-built tool delivered through the NPCC Digital Public Contact framework — and identified 21% more hidden vulnerability indicators as a result. Humberside, Thames Valley and Hampshire deployed Bobbi, an AI assistant that handles 75 to 80% of non-emergency enquiries without human intervention, freeing staff for high-trauma contacts. Kent Police was rated Outstanding by HMICFRS for control room performance after deploying SME triage support tools.

Behind every percentage point is a caller whose risk was identified, or a call handler who had the capacity to do their job properly.

Case studies

Written-up examples from UK police forces, with named organisations and measurable outcomes.

West Yorkshire Police Post Call Analysis — Speech-to-Text AI (SME / NPCC DPC framework)
What they did

AI transcription of 999 and 101 calls scanning audio in real time for vulnerability indicators — specific language, acoustic stress cues, and patterns suggesting escalating risk — that standard operator notes do not capture. SME-built, trained exclusively on police data, delivered within a police-controlled environment via the NPCC Digital Public Contact framework.

Outcome

21% more indicators of hidden vulnerability identified compared to standard manual recording during the pilot. Higher accuracy in THRIVE assessments. Automated compliance checking confirming mandatory advice was given.

Humberside, Thames Valley and Hampshire Police Bobbi (AI conversational assistant)
What they did

Conversational AI integrated into the control room knowledge base to triage non-emergency and non-policing enquiries — parking, general information, low-complexity reports — without requiring a human call handler.

Outcome

75 to 80% of non-emergency enquiries handled without human intervention. Operators freed for urgent, complex, and high-trauma contacts.

Kent Police SME triage support tools
What they did

Risk-grading and vulnerability identification tools deployed to address historic inconsistency in 101 caller assessment and improve outcomes for vulnerable callers at first contact.

Outcome

Rated Outstanding by HMICFRS for control room performance and victim outcomes — one of the most credible independent verifications available in UK policing.

Also worth reading

Further examples from the Proven Public database — not yet written up in full.

South Yorkshire Police
SmartContact AI (NEC)
Automated repeat caller identification surfacing domestic abuse history to operators instantly at point of call.
National — 43 forces
Police.AI Hub
Targeted to free 6 million officer hours annually via automated 101 transcription. Governed by NPCC AI Covenant (2023) and the £115m Police.AI centre (2026). Projected outcome — not yet verified.