Ordnance Survey CTO on AI and the Future of Geospatial Data

The geospatial market is moving away from static images and toward live, intelligent data. 

Whether routing an ambulance or running a delivery fleet, organisations need a constant, live feed of exactly what’s happening on the ground.

Ordnance Survey (OS) has been the steward of this landscape since 1791. While globally renowned for its iconic paper maps, the organisation has transformed over more than 230 years into one of the world’s most sophisticated data providers. 

Today, OS manages the National Geographic Database, a digital engine that processes 30,000 updates daily to maintain centimetre-level precision across Great Britain.

As AI and machine learning redefine how we capture and interpret the earth’s surface, the challenge for technical leadership is balancing this rapid automation with the absolute integrity required of a national service. It is a mission that requires orchestrating a complex pipeline of computer vision, agentic AI and ethical governance to ensure data remains both accessible and authoritative.

Here, Manish Jethwa, Chief Technology Officer at Ordnance Survey, discusses how the organisation is utilising AI to turn 600 million geographic features into actionable intelligence, ensuring Britain’s foundational data stays as resilient as its heritage.

Outline your role and key responsibilities as Chief Technology Officer at Ordnance Survey.

My role is to define and deliver the technology strategy that underpins our position as Britain’s national mapping service. This spans our data platforms, AI and machine learning (ML) capabilities and the infrastructure needed to capture, process and distribute geospatial data at national scale.

A big part of that is evolving the OS National Geographic Database: how we ingest imagery and sensor data, apply automation through computer vision and ML and deliver it to customers via APIs and digital services.

But it’s not just about the technology itself. I spend a lot of time on governance, making sure our AI practices are secure and ethical, that data integrity is maintained and that our hybrid infrastructure stays resilient. 

People are equally central to success here, and balancing innovation with responsibility and creativity is crucial. I want to make sure our teams have the skills and confidence to adopt new technologies effectively.

Ultimately, the goal is simple: ensuring that OS continues to deliver trusted, high-quality location data that works for government, businesses and the public.

Ordnance Survey is known for its maps, but from a data perspective, what is the core value proposition and service OS provides to its diverse customer base?

OS was founded in 1791. Most people know us for maps, and that heritage is something we’re proud of. But when you’ve been around for over 230 years, transformation isn’t new to us. 

What started in paper cartography has evolved into something quite different: we are now one of the oldest and most continuously evolving data providers in the country.

At the heart of that is the OS National Geographic Database, a continuously updated digital master map of Great Britain. It holds around 600 million features, from postcodes and address data, transport networks and building features to land features – updated 30,000 times a day. That data underpins everything from emergency response and infrastructure planning to environmental monitoring, all at centimetre-level precision.

Through our APIs, customers across government, utilities, telecoms, financial services, logistics and beyond can access and build on that data directly. A local authority might use it to plan new housing developments. An insurer to assess flood risk. An emergency service to route resources in real time. 

The common thread is accurate, current location data at the exact moment it’s needed.

AI is changing geospatial data from a specialist resource into an intelligent decision layer. Can you elaborate on how AI, computer vision and agentic AI are practically automating the creation and updating of national-scale geospatial data to achieve this?

AI is fundamentally changing the way we work, and in many ways, we’ve been preparing for this moment for some time. Computer vision and ML have been embedded in our production workflows for over a decade, particularly for extracting features from aerial imagery. 

Tasks that once required significant manual effort can now be handled automatically: models detect, classify and extract features at scale, letting us process national datasets far more efficiently and update them far more often.

With the rise of large language models (LLMs), how is this technology enabling non-specialists to access and interpret complex location intelligence, and what does this shift mean for data usability across different industries?

LLMs lower the barriers to entry significantly. Historically, working with geospatial data required specialist tools and expertise in geographic information systems. Now, users can ask questions in plain language and get meaningful answers from complex datasets.

We see LLMs turning our platforms into something more conversational, letting users in sectors like retail, logistics, healthcare and government interrogate data directly, without needing technical training. 

It shifts geospatial data from a niche analytical resource into a mainstream business tool where users can generate insights, model potential scenarios and plan ahead.

That said, it’s important to recognise that LLMs are an interface layer, not a replacement for what sits underneath it. The real value of AI and LLMs still depends on robust data pipelines and ML models doing the heavy analytical work.

As AI becomes more deeply embedded in geospatial decision-making, why do confidence scoring, robust governance and human oversight remain essential, and what steps is OS taking to ensure data quality and ethical usage?

As AI scales up, trust becomes the critical factor. Automated systems can process enormous volumes of data, but without proper checks they can also propagate errors at the same speed and scale. That’s a significant risk when you’re maintaining authoritative national datasets.

At OS, we tackle this through multi-layered validation that combines automated checks with human review. We apply confidence scoring to AI-derived data so that users understand not just what the system concluded, but how certain it was. 

We document provenance and make uncertainties visible, so customers know where to apply caution. These checks are a critical part in ensuring that data is not only useful, but transparent and dependable – all of which are essential in making decisions at a national scale.

The broader principle is that AI should be a force multiplier for human expertise, not a replacement for it. It reduces repetitive work and surfaces the cases that genuinely need human judgement. 

This combination of automation, validation and oversight is what lets us scale without sacrificing the reliability our partners depend on.

Looking ahead, what is the most significant, long-term impact that agentic AI will have on the business models behind geospatial data in the next 3-5 years?

The biggest shift will be from data delivery to outcome delivery. Agentic AI means geospatial data can become an active participant in workflows rather than a passive input. 

Agents will query datasets, interpret results and trigger actions across systems. Instead of selling access to data, organisations will increasingly offer integrated, intelligent services where value is measured by outcomes rather than datasets.

This creates real requirements around infrastructure. Systems need to be API-driven, secure and built for machine-to-machine interaction. And as agents take on broader capabilities, governance and access control become more important, not less.

Geospatial data is already a critical foundation layer in confident decision making. 

In the next few years, we expect its value to continue to grow, becoming even more impactful within autonomous decision systems. Through continuous, AI-powered interaction, geospatial data will drive smarter cities, infrastructure and citizen-led services.