What happened
Thames Valley Police and Hampshire and Isle of Wight Constabulary reported at Dreamforce 2026 that they have deployed an named Bobbi to manage public-facing, non-emergency contact. The agent, live since November of the previous year, automates approximately 75-76% of routine inquiries while escalating the remaining 25% to human operators. The system uses 1,900 knowledge-base articles to detect topics like domestic abuse or child vulnerability, adjusting its tone and language accordingly to ensure appropriate support and escalation.
Thames Valley Police and Hampshire and Isle of Wight Constabulary shared details of their , Bobbi, at Dreamforce 2026. The agent was deployed to handle non-emergency public contact, aiming to reduce pressure on call handlers and identify vulnerable individuals. Tom Kempster, Director of Digital for the forces, stated that the goal was to allow people to self-serve on common enquiries while ensuring that high-harm situations were escalated to human operators.
Bobbi has been live since November of the previous year, with just over six months of data available at the time of the report. The forces reported that around 75-76% of contacts were subject to automation, while the remaining 25% were escalated to operators on the 24/7 digital desk. The team emphasized that the critical metric is not just the automation rate, but ensuring that the 'right 25%' of contacts—those requiring human support—are correctly identified and escalated.
The agent relies on 1,900 knowledge-base articles covering topics from parking to serious assault. When specific topics like domestic abuse are identified, the agent is instructed to adopt a supportive tone. If language suggests the user is a child, the agent adapts its language to a reading age of ten. The forces tested approximately 3,000 conversations before deployment and continue to monitor and update the agent's behavior, treating it as a 'digital member of staff' with clear accountability and governance structures.
The forces cited a specific example where Bobbi identified a victim of 'cuckooing' (where a home is taken over by criminals). The agent provided information and support over several days, enabling the individual to speak to a human agent. This led to officers being dispatched, four arrests for drug possession, and the safeguarding of the vulnerable person. The forces noted that without the AI interaction, they might not have identified the situation.
Why it matters
This deployment represents a significant practical application of AI in public safety, demonstrating how large language models can be used to triage sensitive interactions without removing human oversight. By automating routine demand, the forces reduce wait times for those in genuine need, while the agent's ability to identify vulnerability markers ensures that high-harm situations are flagged for immediate human intervention. This approach addresses the challenge of balancing efficiency with the nuanced care required in policing, offering a model for other organizations seeking to integrate AI into sensitive service environments.
This deployment illustrates a practical framework for integrating AI into sensitive public services. By automating routine inquiries, the forces free up human resources for complex cases, potentially reducing wait times and improving customer satisfaction. The system's ability to detect vulnerability markers and adjust its interaction style demonstrates the nuanced application of large language models in contexts where tone and empathy are critical.
The governance approach, treating the as a staff member with defined accountability and continuous monitoring, offers a model for responsible AI deployment. This addresses common concerns about AI in policing by ensuring that human oversight remains central, particularly for high-harm situations. The forces' focus on 'trust and confidence' as the true return on investment highlights the importance of public perception and ethical implementation in AI-driven public services.
The case study provides concrete evidence that AI can enhance, rather than replace, human judgment in policing. By identifying vulnerabilities that might otherwise go unnoticed, the agent contributes to preventative safeguarding. This has broader implications for how other public sector organizations might approach AI adoption, emphasizing the need for careful design, testing, and ongoing evaluation to ensure that technology serves human needs effectively.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
What to watch next
The forces are considering expanding Bobbi's capabilities to take actions on behalf of users, such as contacting other departments or individuals. They are also exploring internal agents for staff knowledge access and investigating whether consistent identification of threat and harm can inform preventative safeguarding measures before incidents occur.
The forces are exploring the next phase of Bobbi's capabilities, including the ability to take actions on behalf of users, such as contacting other departments or individuals. This expansion could significantly increase the agent's utility but will require careful consideration of privacy, consent, and potential risks associated with automated actions.
There is also interest in developing internal agents to help staff access knowledge in natural language, which could improve operational efficiency within the forces. Additionally, the team is investigating whether the consistent identification of threat, harm, and risk by the agent can inform preventative measures before incidents occur, potentially shifting the focus from reactive to proactive policing.
The long-term success of this deployment will depend on the forces' ability to maintain public trust and ensure that the AI system continues to perform reliably and ethically. Ongoing monitoring, iterative improvement, and clear communication with the public will be essential to sustaining the benefits of this initiative.