BeeMyGuide

How it works

From a public record to a voice in your ear.

Every place BeeMyGuide talks about starts as rows in open datasets. Here's what happens between there and your headphones.

The pipeline

  1. Gather

    Open, public and openly-licensed sources are pulled for a region: landmarks, monuments, parks, buildings, plaques, the works. Nothing scraped from behind a paywall, nothing bought from a broker.

  2. Match and deduplicate

    The same place arrives five times under four names, sometimes with coordinates that disagree by a city block. They get matched to one another and collapsed into one record, and a call gets made about which version of each fact wins. This is the step everything else depends on, and the one that takes the most work.

  3. Filter

    Plenty of what comes back isn't worth stopping for: a two-sentence stub, a utility box someone tagged, a place that's really just a name on a map. Those are kept in the catalog but held back from the app rather than padded out into something that wastes two minutes of your walk.

  4. Write

    What survives is turned into a short, spoken-word script, written to be listened to while moving, not read off a placard. Lengths vary by how much there is to say and how long you've told the app you want to listen for.

  5. Speak

    The script is turned into natural speech in your language. Twelve languages are live, and the narration voice is the same one that tells you when you've walked out of covered ground.

  6. Cache and serve

    The finished audio and images are stored once and served to everyone after that. The first person to hear about a place pays to create it; the thousandth is reading from a file that already exists. That's the economics of the whole thing, and the numbers are on the FAQ.

Where the data comes from

All of it is public or openly licensed, and the sources are credited in the app.

Wikipedia

The narrative backbone: what a place is, when it went up, what happened there, who cared.

Wikidata

The structured spine: identifiers, coordinates, types, dates and the cross-links that let two records be recognised as one place.

OpenStreetMap

Ground truth for geography: where a thing actually sits, its footprint, and the everyday features that never make an encyclopedia.

National Park Service

Official material for parks, historic sites and monuments, including the details only the people who run a place would know.

Wikimedia Commons

Freely licensed photographs, with attribution carried through to the image you see on screen.

Openverse & openly-licensed Flickr

More photography, filtered to licences that permit reuse, and checked before it gets attached to a place.

Sources are cited in-app, and every image links back to its original so you can see where it came from and who made it.

And the people walking around

Not everything worth hearing is in a dataset. Anyone can build a tour in the tour builder, dropping stops on a map, ordering the route, writing notes in your own words, and the good ones get reviewed and shared with everybody else. A local who knows which door the band came out of beats an encyclopedia entry every time.

Bring me data

I'm actively working on partnerships to get more data into the system: cities and tourism boards, museums, historical societies, park services, transit agencies, universities, archives, tour operators with material sitting in a drawer.

If you hold something, whether a catalog of sites, a set of scripts, an archive of photographs, a walking tour that only exists as a PDF, I'd like to talk about getting it in front of people who are standing right outside it. The bar is that the material is yours to share and that it's genuinely worth someone's two minutes.

Reach out: hello@beemyguide.travel

What's coming next →