In May 2026, the flespi team focused on internal refactoring, adapting the platform to modern AI trends and laying the foundation for future functionality enhancements. The most significant milestone for us was the final release of the so-called “settings name” feature, which allows the same setting/command name to be used with different schemas across different device types. Our lead developer, Alexander Kiranov, worked on this refactoring for the entire nine months, and the result was finally released and deployed in May.
The newly released feature solved the problem of having to design different setting names for each protocol whenever the schema differed. But it also introduced a new question – what should we do with all the existing settings that have somewhat unusual names, such as setdigout_4 or idling_dout3? Should we leave them as they are and accept a bit of inconsistency, or start another round of refactoring across protocols with numerous, and potentially painful, per-device changes for our users? So far, we do not have an answer to this question.
In May, flespi's uptime was 99.9973%, with a few intermittent outages totaling 73 seconds.
One particularly interesting article in our recent flespi team overview series that I recommend reading is about our Support and Marketing. It complements the frontend and backend team articles, with just one final article left to complete the series.
We integrated three new protocols: skybitz-push, magicyo, and teltonika-gateway-wifi-mqtt (wirepass).
For all persistent flespi entities available via the REST API, we added a created field, allowing you to quickly see when a particular entity was created.
Tacho Bridge App 0.7.2 was released with numerous fixes and improvements. In parallel, we also released three major versions of the TachoBox application, adding a wide range of features – from localization into various languages to visualization of border crossings, vehicle inspections, and many other useful insights that we (and of course our users) can extract from tachograph .ddd files processed by the tacho-file-parse plugin.
And finally, one nice niche feature is the ability to enable or disable realms and their users.
The majority of protocol engineering work is now done by AI, while another AI handles 98% of our support and consulting services. This allows our engineering team to focus on much more significant initiatives such as providing an OAuth2 server to connect flespi directly to Claude, ChatGPT, and other AI agentic platforms via MCP, supporting WebRTC video streaming with bidirectional audio, adding support for tacho card hotels, developing an integrated and safe AI agent to simplify daily flespi operations, and much more.
Enjoy the summer (or winter, depending on your location)!