AI Projects – Aspirations, Potential, and Questions

Everyone is talking about AI. But resource-pressured police forces are understandably hesitant to commit to projects that tend towards higher risk, with no guarantee of a useful, usable outcome. You want to reap the benefits of AI but there are a lot of questions. What if the AI ‘hallucinates’? How are decisions influenced by AI data still accountable? Most of us already know the irritation of poorly-aimed AI bots taking the place of good customer service, so we know it’s possible to get it wrong. Poorly trained or managed AI can be a far cry from the most lofty promises of the AI revolution, and for policing, public safety is too important to leave to a chance AI blunder. This 2025 article from The Times describes some of the common concerns for AI adoption.
And yet, the possibilities for policing posed by AI are vast and compelling.
AI Acumen and Innovation
At Bluestar, we offer a smart alternative to ‘the big AI gamble’. We combine seasoned expertise with close agile partnership to help you succeed with your AI aspirations, building realistically from the ground up. Our in-house team has been exploring and proving AI possibilities for several years, from machine learning algorithms built into our products to LLM integration with major cloud providers. Read about our low-risk AI Development As A Service (AI DAAS) below.
We also enjoy substantial backing and support from our parent company Jonas Computing, so we have recourse to a plethora of stablemate companies and expertise, helping us to push forward the very best ideas – like our pioneering work with CHLoE – the AI helper for police investigations.

AI Development As A Service (AI DAAS)
Low risk with tangible benefits
At Bluestar we have a great track record in working closely with Police organisations, delivering innovative solutions for over 25 years. Our approach to AI is underpinned by that proven success. We’re with you throughout your journey to leverage AI from early inception right up to the development of carefully optimised production systems. Our AI Development As A Service (AI DAAS) model gives you access to key AI expertise within an agile, iterative development model, so you get early access to prototypes as development progresses, allowing you to steer the project and get the best possible outcome.
We started our AI journey some years ago and can boast success with a wide range of techniques. This has come from the right combination of:
- First class expertise – in machine learning algorithms, textual analysis and LLMs (Large Language Models).
- Targeted integration – harnessing AI in ways that enhance an overall human process. You stay in control.
- Data know-how – leveraging our expertise, combining and processing RMS data and more, in our core platform, Corvus.
For examples of successful AI outcomes, look no further than our products below or ask us about AI Development As A Service.
AI Products
Corvus Helper for Lines Of Enquiry (CHLoE)*
Beginning years ago with a single customer’s need to analyse thousands of non-incident 101 calls, our LLM R&D expertise is taking the next big leap. We’re now developing AI-driven, investigation processes with CHLoE – our AI for investigation teams. CHLoE is designed to prioritise lines of enquiry, advising on solvability and steering the actions most likely to yield a result. CHLoE aims to reduce the strain on many investigation teams by reducing painstaking activities. CHLoE is designed to be trained on years worth of investigation history, fine-tuning its ability to guide today’s investigations in the right direction.
*CHLoE is under development.
Corvus Crime Summaries
Officers don’t always provide a good modus operandi on crime records. Sometimes the MO information is subjective; sometimes a temporary submission with the intention of completion later. But the officer often finds they don’t have the time to revisit and revise. And for most investigation teams, the initial description of an incident is quickly usurped by information added later, as the investigation progresses. The result is poor MO data.
An ancestor of CHLoE, Corvus Crime Summaries uses LLM AI technology to analyse the content of Investigation records. Initially aimed at automating adjustments to MO data, the capability was expanded. Corvus Crime Summaries now provides a succinct summary useful to all stakeholders by ‘trawling’ the RMS information throughout the Investigation record and its related records. There is also a configurable ‘write back’ option, to update your RMS.
Corvus Call Insights (CCI)
Our proven Call Insights solution uses a trained AI Large Language Model to transcribe, analyse and categorise thousands of call recordings, revealing the trends that put strain on control staff and resources. Our reference Force found that around 50% of calls for service managed by the Force Contact Centre were not resulting in a formalised C&C system log being created. Yet the calls still placed a significant burden on call takers.
Corvus Call Insights sheds light on all calls so you can help your staff by reducing their burden, by revealing the ‘themes’ or ‘topics’ that frequently come up. CCI uses ‘topic modelling’ – an LLM AI analysis process that can be precisely tuned to take account of common misunderstandings, resulting in highly accurate analysis of the load placed on call handlers by miscellaneous and non-incident calls. The algorithm can be precisely tuned, even catering for local colloquialisms (for example ’roundabout’ can be an expression signalling a time period, rather than relating to an RTC!). Call Insights can run against a backlog of call recordings, but also runs live – against today’s incoming calls – feeding a live dashboard showing all current call handler topics, so if there is a sudden change and increase in demand, you can act quickly.
The overall result is an enhanced visibility of all calls and numerous opportunities to reduce the strain on call handlers, improving their morale and wellbeing.
Here’s an overview of the LLM ‘topic modelling’ process:

Ask us a question or watch our in-depth webinar that explains the innovative AI-driven processes and the realised benefits for one partner force.
Machine Learning Products
Before LLMs like ChatGPT and Microsoft’s CoPilot became popular, we were already delivering AI value by utilising Machine Learning algorithms. We now have a depth of expertise and referenceable AI solutions, delivering value without disruption.
Here are some of our success stories:
Custody Cell Allocation Management System (CAMS)
Our trailblazing solution directs arrests to the right custody suite based on location, availability of cells, specialist devices (e.g. a Camic machine) and predicted occupancy, all using near real-time live data. Under the bonnet, CAMS uses the Genetic ‘cross breeding’ AI Machine Learning algorithm to rapidly examine millions of possibilities before directing the arresting officer. The result is fewer redirections to other suites, less delay at the point of arrest and a smoother booking-in process. Read more about CAMS here.
TreadMatch & The National Footwear Database (NFD)
We developed the National Footwear Database which uses our pattern matching algorithms to match footwear impressions at crime scenes against those taken in custody – and much more. Read more about the NFD and TreadMatch here.
Data To Feed Your AI Projects
At Bluestar we don’t simply offer AI-powered solutions. We can also offer componentry to help you utilise data within your in-house AI projects.
Bluestar Police data know-how, feeding Cloud AI
For UK policing, there’s no denying the possibilities being offered by big cloud providers like Microsoft Azure. The emerging trend seen towards cloud provider LLM AI product sets partly owes itself to the compelling way cloud providers are bundling sets of related capabilities together, and Police Forces are keen to reap the benefits of familiar LLM AIs like Microsoft CoPilot, orchestrating data through Microsoft’s Fabric platform, producing reports and dashboards with Microsoft Power BI etc. But questions remain about how to surface data required to feed these initiatives. This is where we come in.
At Bluestar we can provide the technical ‘building blocks’ and APIs to support your LLMs. We have best-in-class experience in aggregating data from across your organisation, giving you a singular consistent data model combining RMS, C&C, HR and specialist system data. We also provide application capabilities such as mapping and visualisation. With our LLM AI expertise and detailed police domain-modelling and integration experience, we’re uniquely positioned to support your Force’s journey into AI from ‘underneath’ – providing the data platform to feed your AI projects.
Example
A CoPilot LLM trained in understanding the analytics requirements and language of your Police Force could:
- Utilise a singular aggregated Data API to access your Corvus data superset, representing your C&C, RMS and HR systems, augmented with numerous other data sources
- Provide output either through a Geographical Mapping user interface or to an in-house reporting function
At Bluestar we offer the data integration that police organisations need, to underpin rich AI functionality. Contact us to discuss your requirements with a technical expert.
