Case study
AI-generated audio tours, built to scale
Featured in Deutsche Telekom's Magenta Moments
Guidable builds a personalized audio tour from a visitor's interests and location: retrieval-augmented generation in Python over a vector database of city content, not a flat list of nearby points. We built the generation, the discovery layer, and the playback engine, for hundreds of thousands of users across Germany, with content spanning 500+ cities in 40+ languages.
At a glance
The engineering
The hard part was not streaming audio
An audio-tour app looks simple until it meets a real city, a real network, and real users on the move. Three problems decided whether a tour felt seamless or broke mid-walk.
Discovery happens before playback
A visitor in a new city needs the right tour before they start walking. We built the discovery layer to surface tours by proximity and theme, not as a flat list, so the right walk is one tap away.
Audio that never pauses mid-tour
The silent killer was buffering on slower mobile connections. We structured content delivery so the next segment loads in the background while the current one plays, and the listener never hears a gap.
State that survives the real world
Variable connections, a phone call, a locked screen, the app backgrounded. The playback layer handles background-mode interruptions and recovers mid-tour state, so a walk picks up exactly where it left off.
Scope
What we built
The product
Guidable in the wild
Built with
The stack
Where it landed
FAQ
Questions buyers actually ask
Who built Guidable?
What does Guidable's AI tour generation involve?
Let's build yours
Our team contacts you within 24 business hours
We collect all the key requirements from you
The team of developers prepares estimation
We can sign NDA since we respect the confidentiality of our clients