IT/OT Builds · In partnership with HiveMQ
We put HiveMQ’s new platform on our shop floor. Here’s what happened.
Four videos, one bakery, no cuts (to be honnest: one fast forward). HiveMQ’s Shashank Sharma and Kudzai Manditereza join us to explain their move from products to platform, then build it end to end on Sweet Harmony Treats.
The case
Sweet Harmony Treats
Sweet Harmony Treats is the bakery we use to make these things concrete in our Academy and now also in our IT/OT Builds. Multiple sites at wildly different levels of maturity, PLCs and HMIs from a dozen brands, the legacy SCADA project bolted on wherever a plant manager needed it. No single team owns “data.” Nobody agrees who should.
That combination is common, and it usually breaks in the same order. Connectivity first: the data simply isn’t reachable yet. Then, once it’s flowing, the harder wall: does “oven temperature” mean the same thing on line 3 in Munich as it does on line 1 on the other side of the world?
That’s the problem we asked HiveMQ to solve in front of us.
The architecture
From products to a platform
HiveMQ Edge and the Enterprise Broker are the existing Connect layer. What’s new are three layers built on top of it: Contextualize (HiveMQ Pulse, now fully absorbed into the platform rather than sold separately), Analyze (a data-trust and health score with deviation alerts), and Act (a governed, agentic layer).
Contextualisation happens at the edge, but is managed through their platform.
Where this fits
Where this sits on our Capability Map
We don’t just host the pitch, we place it. Here’s where HiveMQ’s platform lands on our Industrial Data Platform Capability Map.
- 1 – Connectivity is what HiveMQ also calls Connect. It’s covered in the first technical tutorial.
- 2 – Transform, Contextualize & Structure and 5 – Edge Compute come together in what HiveMQ calls Context, shown in the second tutorial. Note that this covers part of how we define Edge Compute, not its full scope.
- 7 – Data Sharing isn’t covered in the tutorials yet, but it’s discussed in the platform overview when we talk about Agentic AI, what HiveMQ calls Act.
The build, part one
A full platform walkthrough
Kudzai walks us through the new HiveMQ platform step by step
The agentic “Act” layer shown here is public preview, not GA yet.
The build, part two
Two tutorials, follow along
These are hands-on, and the videos are uncut by design. Feel free to follow us along on your own instance.
Connect
Open in YouTube →Deploying HiveMQ at the edge and moving from ungoverned traffic to a typed, governed namespace — sites, lines, assets, metadata.
Context
Open in YouTube →Reusable data models, calculations, writing back to our SCADA system (or any other data consumer) and more.
What you’d need to follow along: a free HiveMQ account (100 connections). If you want to Vibe Code with us, you'll need Claude Code and Docker as well. Everything shown is something you can do on your own machine today.