Hardware ships. Software decides whether it survives.
From prototype to production, Naxtre develops IoT solutions for smart meters, cold chain sensors, EV chargers, payment terminals, safety devices, smart translation glasses, and grid transformers and many more. We create firmware, provisioning systems, cloud platforms, and user friendly dashboards and apps.
Most agencies build the app. We build everything the app is lying about.
A connected product fails in the gaps: a safety device whose battery is flat when it is finally needed, a payment box production line that cannot scale without the right firmware and identity on every unit, a faulty meter holding readings no user can reach, a transformer platform that buckles as data volume grows. Sabre, Paytm, Oakmeter and Hitachi are where our last decade has been spent, closing exactly those gaps.
Software that rescued the hardware. Sabre's personal safety device was sound in principle but draining its battery in the field. Rather than touch the hardware, we rewrote the device software and connectivity logic and cut power consumption by up to 90 percent, the difference between a device that is charged when it matters and one that is dead in a drawer.
Production at scale, cost down. For Paytm Soundbox we built the systems that track manufacturing status, map QR codes to IMEI numbers, load the right language firmware and clear a unit for dispatch, at 99.9 percent accuracy and a fraction of the manual cost.
Industrial grade when it counts. Hitachi Energy brought us in to replace ageing transformer services with a scalable platform covering health monitoring, predictive maintenance and real time analytics across their renewable fleet.
Portable by design. Tessol needed temperature telemetry inside other people's apps, so we shipped a platform agnostic short code that drops into any mobile stack without dragging performance down.
Five layers, and we are accountable for all of them.
Bring us a board that already works, a product still on the whiteboard, or a fleet you inherited from someone else. We pick up at whichever layer is broken.
Embedded and firmware
Power budgets, duty cycles, kernel level audio and sensor tuning. The work that decides whether a device lasts a week or a year on a charge.
- Power optimisation
- Android kernel work
- Sensor calibration
Connectivity and provisioning
Pairing flows that hold up on real networks, secure onboarding, over the air updates and fleet grouping that does not fall apart at scale.
- BLE · Wi-Fi · NB-IoT
- OTA update pipelines
- Zero touch provisioning
Platform and cloud
Ingestion, device registry, rules, storage and APIs built to absorb growth in message volume rather than buckle under it.
- AWS · Azure
- Device registry
- Event pipelines
Control apps
The interface your customer judges the product by. Scheduling, sharing, notifications, activity history and offline behaviour that makes sense.
- Flutter · React Native
- Native iOS · Android
- Arabic and full RTL
Data and predictive maintenance
Turning raw telemetry into an operations view: asset health, anomaly flags, consumption analytics and reporting leadership will actually open.
- Power BI
- Health scoring
- Anomaly detection
One team, one accountable partner, many IoT solutions.
Every project here is a product in the field, not a concept deck. Browse by category, open a project, and read the brief, our part in it and what changed.
The brief
Oakter needed a single app to configure and control an entire product family: OakPlug smart plugs, the OakView Wi-Fi camera, OakRemote and the Oakter hub. Every device had its own pairing behaviour and its own idea of what a control screen should be.
What we did
We folded all of it into one platform with scheduling, auto off timers, remote control from anywhere, shared access for up to five users, per appliance notifications and a full activity history. The existing architecture would not scale, so it was rebuilt underneath the interface.
What changed
One app across the full Oakter smart home range, with an interface simple enough that installation videos sit inside the flow rather than in a support portal.
Project visual
Reading the stream is easy, Acting on it is the work.
Every device we have shipped produces more data than anyone watches: temperature every few seconds from a Tessol probe, load from an Oakmeter feeder, health signals from a Hitachi transformer, a payment landing on a Paytm box. The value appears the moment something reads that stream and acts before a person would have noticed anything was wrong.
Predictive maintenance on the grid
For Hitachi Energy we replaced transformer services that could not carry growing data volumes with a platform built for health monitoring and predictive maintenance. A model watches the signals an engineer would watch, on every asset, without ever looking away.
Breach caught while the load is saveable
A Tessol probe that only logs is evidence after the loss. One that recognises a drift pattern flags the freezer while the food or pharma inside can still be saved. The reading matters less than the moment it is acted on.
Meaning from ordinary meter data
Consumption, load and duty cycle from an Oakmeter feeder carry more than a number on a bill. Learned over time, those signals surface tampering, forecast demand and flag a fault before it becomes an outage.
Inference on the device
Some decisions cannot wait for a round trip to the cloud, and plenty of sites have coverage that would never carry a full data stream anyway. Running the model on the hardware means the device sends its conclusion rather than its raw feed.
Ask the fleet a question
Nobody should have to learn a dashboard to find out which units are struggling. A conversational layer over the same telemetry answers in Arabic or English, in plain language, and escalates on its own when the answer is a problem.
Systems that tune themselves
Loads shift, seasons change and setpoints drift. Models that adjust charge scheduling, sampling rates or cooling in response cut energy and wear without anyone opening a control panel.
Four stages. You can stop after any of them.
No twelve month commitment required to find out whether we are any good. Each stage ends with something you can run, test or take to your board, and a straight decision about whether the next one is worth doing.
Technical discovery
Hardware review, connectivity assessment, power budget, and an honest read on what the current stack can and cannot carry.
Working prototype
One device, one end to end path. Pairing, telemetry, cloud, app screen. Real hardware in hand, no simulator theatre.
Pilot fleet
Fifty to five hundred units in real conditions. Provisioning at volume, OTA rollout, failure modes documented and closed.
Scale and support
Production rollout with a dedicated team, release cadence, monitoring, and factory line tooling as volumes grow.
Tell us what the device has to do.
Send the product, the problem and the deadline. An engineer replies, not a sales sequence. If we are the wrong fit we will say so in the first call.