Air Aware
Web app that provides real-time air quality data for the London Boroughs using AI agents.
Real-time air quality data the public could not act on.
Four London boroughs (Hackney, City of London, Tower Hamlets, and Newham) were publishing real-time air quality data. Levels of PM2.5, PM10, and nitrogen dioxide regularly exceeded Air Quality Objectives. Residents had no easy way to interpret the readings or know what to do on a high-pollution day.
Static dashboards and PDF guidance were not landing with the people most affected. Typical government digital services see sixteen-second average visits and 20-30% scroll depth, which meant vital public-health information about pollution simply was not reaching residents.
The boroughs needed a single front door for residents: multilingual, available around the clock, accessible on older devices, and grounded in their own published data and public-health playbook, not a generic chatbot.
- Public-facing service. No specialist knowledge required to use it.
- Answers must be grounded in the boroughs' published data and guidance.
- Available in the languages each borough already serves.
- WCAG 2.1 AA accessibility end-to-end.
- Works on low-bandwidth and older devices.
Four reasons the borough's existing channels weren't reaching residents.
Complex data was inaccessible
Air quality data sat in technical reports and government databases that residents could not understand or use to make daily decisions.
Government engagement was thin
Typical government digital services see sixteen-second average visits at 20-30% scroll depth. Critical health information was not reaching the people who needed it most.
No personalised advice
Residents had no way to get tailored recommendations based on location, health conditions, or daily routines. Generic guidance, broad-brush at best.
Data without action
Even where residents accessed pollution data, there was no clear path from a reading to a step they could take to protect their family that day.
Inside the Air Aware service.
Conversational answers grounded in the borough's own data
The assistant explains the current air quality reading in everyday language and offers a clear next step. Every answer cites the borough's published data and public-health guidance. Handles natural-language requests like 'when is air pollution lowest in Hackney?' or 'how did air quality change in Newham over Christmas?'
Real-time and historical pollution data, mapped
Interactive map combining DEFRA monitoring stations and local sensors across the participating boroughs. Drill down by area, compare historical trends, and understand the impact of weather on local air quality.
Health and exposure guidance in plain language
Articles, practical advice, and educational content curated from DEFRA, NHS, the British Heart Foundation, and peer-reviewed research. Step-by-step guides for protecting vulnerable family members, translated to match each borough's published communications.
Local events, news, and ways to take action
Borough-curated events alongside news aggregated from the BBC, The Guardian, and Air Quality News. Residents see what their council is doing locally and how to engage with community air-quality initiatives.
A public AI service grounded in the boroughs' own data, not a generic chatbot.
We built a public web app and AI assistant on top of the boroughs' real-time air quality feeds. The assistant retrieves the relevant published data and guidance, then answers in plain language with a clear next step. No sign-up, no app install, accessible on older phones.
Design ran iteratively with each borough's public-health and communications teams. Early static mockups progressed to clickable prototypes tested with residents directly. Content was reviewed by each borough's teams before going live. Multilingual support was added during community feedback rather than designed in a vacuum. The platform meets WCAG 2.1 AA standards, with adjustable text sizes, high-contrast settings, and colour-blind friendly design.
Deployment extended beyond the web: physical screens in pharmacies and public spaces brought the same content into real-world community touchpoints.
- Conversational AI assistant grounded in the boroughs' published air quality data.
- DEFRA monitoring-station and local-sensor data integration, plus a real-time pollution map.
- Curated content layer over DEFRA, NHS, British Heart Foundation, and peer-reviewed research.
- Separate CMS for council communications teams to publish events and updates.
- Multilingual content, adjustable text and high-contrast modes, low-bandwidth performance.
- Physical-screen deployments in pharmacies and public spaces.
From static mockup to clickable prototype to the service that shipped.
Two static mockups and a clickable prototype were tested with residents and each borough's communications team before any production build. The interface that shipped reflects what landed in those sessions, not the first concept sketch.
Iterative prototyping meant residents and borough partners shaped the surface in real time. Multilingual support was added during community feedback rather than designed in a vacuum. The interface meets WCAG 2.1 AA throughout: adjustable text sizes, high-contrast settings, colour-blind friendly design.
Average resident session length against the government-website baseline.
Measured in the first two months of operation against the GOV.UK and council-website industry baseline of a sixteen-second visit. Same audience, same environment, comparable measurement window.
Average resident session length (longer is better)
Five-minute sessions, four boroughs, deep public engagement.
Resident engagement
Five-minute average sessions against a sixteen-second government-website baseline. Residents stay long enough to actually use the guidance.
Deep interaction
Residents read the guidance rather than bouncing immediately, far ahead of the typical 20-30% scroll depth on government sites.
Accessibility
WCAG 2.1 AA throughout, with adjustable text sizes, high-contrast settings, and colour-blind friendly design.
Community reach
Live across four London boroughs with physical screens in pharmacies and public spaces, extending digital engagement into the high street.
Borough communications teams publish updates without touching engineering.
Each borough has its own communications and public-health teams already producing updates, events, and local advice. Air Aware ships with a separate workspace where those teams write, edit, and schedule the content that residents see, with no developer in the loop.
- Events & updates. Council teams across all four boroughs publish events and updates that appear chronologically on the public site.
- Built for non-technical authors. Designed for communications teams, not engineers. No CLI, no markdown, no deploys.
- Borough-scoped. Each borough's content stays attributable to that borough. No cross-posting unless they choose to.
- Reviewed before publish. Each borough's editorial workflow runs end-to-end inside the workspace.
Government communications, regulated data, public health.
Air Aware's pattern transfers to any organisation where complex regulated data needs to land with a non-specialist audience. Public health, environmental data, accessibility tooling: the boroughs didn't just improve their digital presence. They changed how citizens interact with critical health information. Where the channel is the bottleneck on outcomes, an AI assistant grounded in published data is a more useful surface than a portal full of charts.
From the outset, their team demonstrated a clear understanding of the brief which required a bespoke solution to be designed incorporating the different requests from a variety of stakeholders. Throughout the project, OpenKit's team remained focused on our objectives and was able to consistently prioritise the needs of our communities. Their thoughtful approach, combined with meticulous attention to detail, made working with them not only effective but also a genuine pleasure.
OpenKit developed an impressive AI assistant, providing an engaging and informative way for our residents to understand air pollution. The tool's user-friendly interface and interactive features have made it easy for people to access the information they need. Despite the challenges faced during the project, OpenKit managed to deliver a product that exceeded our expectations.
Lily Mitrijevaite, Environmental Control Officer at Newham Council How we delivered it.
Stack
Capabilities
Compliance
From scoping to live.
- Stakeholder workshopsJoint sessions with public health, communications, and digital teams across four boroughs. Defining what 'understanding air pollution' had to mean to a resident. Weeks 1-3
- PrototypeAI agent grounded in the boroughs' published air quality data and the public health response playbook. Conversational web prototype tested with residents and iterated through clickable mockups. Weeks 4-8
- Public launchRolled out across the boroughs' communications channels. Multilingual support added during community feedback. Physical screens deployed in pharmacies and public spaces. Weeks 9-20
- Steady-state operationWeekly content updates with each borough's communications team, quarterly model evaluation, annual public health impact report. Ongoing
Read another story
All case studies
Bring your team's next AI project to a 30-minute call.
No deck. We listen, sketch a delivery shape, and tell you honestly whether AI is the right tool for the problem.
