If you work in geospatial, your first reaction to “Overture Maps” was probably some version of: don’t we already have enough maps? OpenStreetMap exists. Google and Microsoft have their own. Cities publish authoritative data. So why does the world need one more?
That’s the question MapScaping Podcast Host Daniel O’Donahue posed to Amy Rose, CTO of Overture Maps Foundation, on a recent episode of The MapScaping Podcast – GERS, Metadata, and Maps Built for Machines. Her answer reframes what Overture is actually trying to do.
How Overture Maps Differs from OpenStreetMap (OSM)?
Many of the organizations and communities building maps are designed around taking contributions. OpenStreetMap (OSM) is the clearest example – a community of people putting things on the map, and a vital part of the open data ecosystem. But that’s a different job from being production-ready out of the box. Every data source, OSM included, has its own schema, its own collection method, its own preprocessing. Amy put it this way:
“Overture is something that exists downstream and is able to bring together all these data sources and standardize them in a way that there’s enough standardization that it can be immediately deployed for production use.”
Overture isn’t trying to replace other sources, Overture collaborates with them. They do the hard work of collecting the data, and Overture takes on the less glamorous job of standardizing messy, inconsistent inputs into something you can actually drop into production.
The operating philosophy is what Amy calls a “light touch” – just enough structure to make the data usable, without trying to be comprehensive or do everything. Overture isn’t aiming to be everything for everybody. It’s aiming to be reliably usable for what it does cover.
Metadata Built for Machines and AI
Historically, metadata has been something everyone wants but nobody wants to actually write. That’s changing fast, because the primary consumers of geospatial data increasingly aren’t humans – they’re machines: models, pipelines, automated systems.
That shift matters because humans are good at noticing when context is missing and filling gaps intuitively. Machines don’t do that by default. They generate a plausible-sounding answer and move on, whether or not it’s grounded in anything real. When the data touches critical infrastructure or has consequences for human safety, that stops being an academic distinction.
For machines to give the right answers, they need context – and that means access to metadata at the record level, not just the dataset level.
Overture builds its pipeline around exactly that. Amy walked through what it looks like in practice:
“We bring in hundreds of data sources into our pipelines to get at this one opinionated release that comes out every month. What that process looks like is: bring in a source dataset, log everything from where it came from to what the dataset is, what version, when we got it – and that information stays bundled with it throughout our pipelines. We run validation checks that also get bundled at each individual attribute level.”
Everything that flows through ends up in what Overture calls the corpus – every record, from every version, of every dataset ever processed, stored and matched across sources.
What is GERS (Global Entity Reference System)?
GERS – the Global Entity Reference System – is the unified system that helps Overture track metadata and enrich data.
In practice, GERS assigns a persistent ID to real-world entities: buildings, roads, places, addresses. The payoff: the same building can carry an identical ID whether you’re pulling it from OpenStreetMap, Google, Microsoft, or a city dataset like Vancouver’s.
Instead of everyone arguing about whose polygon is “correct,” GERS lets different sources agree on identity while still disagreeing on geometry. And because Overture keeps the underlying source records, you’re not locked into Overture’s own choice of polygon – you can trace back and pick a different representation if it fits your use case better.
Every Overture release ships with a Bridge file: here’s what you see in the release, here’s its GERS ID, and here’s that same building’s ID in Google, in Microsoft, in the city of Vancouver’s own dataset. A GERS ID is randomly generated and published as part of the reference map, functioning as a kind of Rosetta stone connecting datasets that otherwise have no reason to agree with each other.
Stability matters here too. Overture maintains a GERS registry and a GERS changelog specifically so IDs persist over time – and when something does change, the changelog is how you navigate that instead of getting blindsided by it.
Why Interoperability Over “One Perfect Map”?
Is there really a single “best” representation of a place – or only representations that are fit for a particular purpose? Amy argues that instead of looking for the one correct version of a feature, you need to ask what representation actually serves the task in front of you.
“Depending on my use case or the purpose I’m using the data for, I might want a very generalized building. You might want a very detailed building with the building parts and the heights and rich attributes. The idea here is not to converge on the same representation, but to recognize that there can be many representations – we just need to agree that we can connect them in some way.”
That’s the actual goal: not sameness, but interoperability through GERS – different organizations building their own best maps, for their own purposes, still able to talk to each other underneath it.
Conclusion: Building the Geospatial Knowledge Layer
This podcast isn’t a conversation about map data as a static, finished product. It’s about infrastructure – the connective tissue that lets different maps, built by different organizations for different reasons, actually talk to each other. GERS is the identity layer. Metadata is the trust layer. And increasingly, both are being built with machines, not just humans, as the end user in mind.
If you’ve already listened to, and liked this podcast episode, there’s plenty more where that came from on the Mapscaping podcast. If you missed this one, the full episode is linked below.
Listen to the full podcast here.
Start Building with Overture Maps
Whether you’re a city planner aligning municipal infrastructure, a mobile developer adding location features, or an enterprise team grounding AI agents in physical reality, Overture gives you a structured base layer to build on. It won’t do everything – we’re the foundation, not the finished product – but it’s a solid place to start.
Come build with us. Here’s how to take the next step:
- Explore Overture’s vision for open spatial data and location grounding.
- Join Overture and help shape the knowledge layer. Visit the Overture website to learn how your organization can become a member.
- Dig into our reference data and system through the developer documentation.
- For the latest schema releases and AI tooling, sign up for the monthly newsletter or follow Overture on LinkedIn, X, and Bluesky.
