TL;DR
On June 10th, Bosch MPS and Zeliot teams delivered more than a webinar, we delivered a live production-grade streaming system, end-to-end, in real time.
In just under 30 minutes, the Condense team built a fully operational data pipeline: ingesting from physical vehicle devices, applying custom and prebuilt stream logic, and routing alerts to live dashboards and structured storage. No mock data. No patchwork setup. Just live Kafka-powered execution.
Watch the recording: Build a Data Pipeline Under 30 Minutes using Condense 🚀
What Makes This Different
This wasn’t a tutorial or marketing demo. It was a working example of how real-time architectures should actually work.
At the center was Condense, a vertically optimized, fully managed Kafka-native platform. But unlike traditional Kafka platforms, Condense brings in:
BYOC Kafka: run in your own AWS, Azure, or GCP, retaining data sovereignty and leveraging cloud credits. The entire pipeline runs inside the customer's own cloud account using Condense's BYOC deployment vehicle data never leaves the enterprise's cloud boundary.
Protocol-aware device ingestion: connectors for iTriangle, Teltonika, Bosch MPS, and more.
GitOps-native custom logic: deploy transforms with Docker, Git, and Kafka bindings from the UI.
Built-in observability: topic health, lag, retries, transform traces, all without external tooling.
Real-world mobility readiness: outputs to PostgreSQL, AquilaTrack, and other operational systems.
End-to-end pipeline observability was live from the first event consumer lag, transform throughput, and sink delivery rates all visible without any external monitoring setup.
What Was Built, Live?
Ingested live data from a demo car fitted with an iTriangle device
Added a Teltonika device to simulate multi-hardware ingestion
Deployed a Panic Alert transform from GitLab using the built-in IDE
Added a Periodic publisher for heartbeat events to AquilaTrack and PostgreSQL
Forwarded alerts to AquilaTrack’s live dashboard via API
Stored structured data in PostgreSQL tables for downstream analytics
And finally, when the actual panic button was pressed in the vehicle, the system lit up:
Kafka decoded the payload in real-time
Custom transform flagged the alert
PostgreSQL recorded it
AquilaTrack showed the event live
Why This Matters?
Vehicle data is transmitted via MQTT from telematics devices into Kafka topics where Condense's stream processing layer transforms raw telemetry into actionable intelligence. This webinar proved that real-time is not just about fast brokers, it’s about fast decisions.
Condense redefines what streaming-first systems can be:
Kafka without complexity
Logic with CI/CD and Git control
Device-to-dashboard observability built in
Domain-ready connectors that eliminate months of plumbing
It wasn’t a simulation. It was a system, running live, built transparently in front of an audience.
Watch the Replay
The same workflow pattern ingest, transform, alert, route can be applied to any real-time data source, not just vehicle telemetry. For anyone building connected mobility platforms, telematics systems, or vehicle intelligence workflows, this is worth watching. Not just for the product, but for the architecture, velocity, and clarity of execution.
For a customer example of this same architecture powering a production fleet intelligence platform at scale, see how iTriangle telematics devices feed real-time fleet decisions. To Learn More, Visit this page: Zeliot x Bosch MPS
