We build the software that drives your business.

The challenge

Our client manufactured lamps sold into offices, shopping centres and other indoor buildings — with no connectivity at all. Maintenance meant sending a team of technicians to each installation on a schedule to look for problems, which was expensive enough to eat visibly into the margin. Meanwhile their own customers were asking for the remote control they had already seen from competitors. Three things had to be solved at once:

• alerting on failures, instead of discovering them on a site visit;
• consumption monitoring precise enough to prove energy efficiency;
• remote control of the lighting system.

What we built

Time to market was the hard constraint: customers were already migrating to competitors with smart lighting. We delivered the full stack in parallel.

• Firmware for the ESP32 control and power-metering device, with the hardware designed together with the client.
• An edge computing application on site, so installations keep working when the link does not.
• A core backend layer implementing the public API and device communication over MQTT.
• A single-page web application for monitoring, control and reporting.

The outcome

A complete IoT solution from the device to the cloud platform, which changed how the client's customers run lighting in large offices and malls. Fine-grained consumption data gave them a tangible way to measure energy efficiency and to start doing predictive maintenance. The cost saving in the first year alone covered the entire investment in building the platform.

The challenge

Industry runs on expensive machines that are often decades behind when it comes to talking to cloud, web and mobile systems. The data inside a factory is valuable, but it is stranded: the devices only speak legacy protocols, on local networks, and replacing them is not an option. Connecting that data to the outside world is exactly where we can help.

What we built

A Python application deployed on a Raspberry Pi acting as an edge computing platform. It talks to Modbus devices over RS‑485 and acts as a gateway to the cloud over MQTT — a standard protocol that integrates with practically anything. Production figures and alerts became available in near real time, from anywhere.

The outcome

The gateway changed what the company could actually control: processes could be watched and reacted to as they happened, not reconstructed afterwards. Because MQTT opened the door to a wide range of platforms, every connected factory optimised its workflow within a short time — a real step forward in efficiency, with no machine replaced.

The challenge

This one started as an internal idea. We were asked to build auction systems often enough to notice a pattern: they were all web applications, and they all felt the same. We wanted to try something else and use a platform we like a lot — Telegram — as the whole interface.

What we built

A Telegram bot that serves both sides of an auction: people publishing items and people bidding on them. Users create an auction from inside Telegram itself, and an approval step sits in front of publication to keep out content that would infringe copyright or other rules. Once approved, an auction gets a start date and time; when it opens, the bot recreates the pressure of a real auction room through the chat itself.

The outcome

A proof of concept that has since become the starting point for a number of custom projects, and that we still improve from time to time. The original bot is open source and you can try it live.

Open @AstoneBot on Telegram    Source on GitHub