[Technology and AI partner]

Run your connected devices with bespoke, reliable, intelligent software

Q: How long does it take to develop a connected application?

A first prototype within a few weeks, then delivery in modules so you can test on your hardware before the final version is even ready. The full timeline depends on scope, which we define with you up front.

Q: Do you also develop the firmware for our devices?

We cover the application software: mobile or web app, data collection and monitoring platform, and models embedded on your hardware. We work with your hardware teams or your manufacturer on the interface between the device and the app.

Q: Can the app communicate with our existing hardware?

Yes. We connect to your device’s protocol, over Bluetooth Low Energy, Wi-Fi or a cellular network, and integrate the platform with your IT systems.

Q: How do you handle connection drops in the field?

We design the app to work offline and resync data as soon as the network comes back, with a local queue and conflict handling rules agreed with you.

Q: Can our data stay on site?

Yes. We run the processing directly on your hardware when needed, as on Castrol’s filling line, where no image ever leaves the plant.

Q: Can we start from an existing application and build on it?

Yes. We take over an application or platform that is already deployed, make its foundations more reliable and get it ready for your next requirements, when that is the most sensible choice for you.

Q: Who owns the work produced?

You do. Code, mock-ups, trained models and documentation belong to you at every delivery. The transfer of rights is written into the contract from day one.

Q: How much does it cost to develop a connected product?

Every project is sized to your scope and priorities. Tell us about your context and we’ll come back to you within 24 hours with an initial assessment.

We design the applications, monitoring platforms and embedded intelligence behind your connected devices. You get field data you can actually use, and users who actually adopt it.

50+ projects delivered for start-ups and manufacturers

Expertise across the whole chain, from hardware to the AI software layers

140+ senior tech specialists trained at Polytechnique, Centrale or ENS, among France’s top schools

Talk to an expert See 4 projects

Oghji – mobile app for controlling the connected electrical panel

50+ manufacturers and connected product makers use software built by our teams.

Two people in front of a laptop showing an AI chat interface

[How we work]

Our method for building your IoT application

SCOPING
Step 01
Hardware constraints and field use
We start from your devices, their communication protocol and their real operating conditions. From there, we define the software scope and size the hardware you need.
DESIGN
Step 02
Testable prototype and interfaces
Our designers create mock-ups of the control and monitoring screens and have your field users sign them off. That way we know the interfaces are easy to read before any development starts.
DEVELOPMENT
Step 03
Development and integration with your IT systems
Our engineers build your solution in regular iterations and integrate it securely and robustly with your existing tools and IT systems. We deliver the communication and data collection layer first, so you can test on your own hardware within the first few weeks.
EMBEDDED INTELLIGENCE
Step 04
Models trained, then embedded
Our AI engineers train models suited to your device’s data (images, sensor signals or sound), then optimise them to run directly on the hardware, with inference times that hold up in real use.
VALIDATION
Step 05
Real-world testing and commissioning
We validate on your devices against the connectivity, battery life and environmental conditions they face in the field. Then we publish the apps, put the platform into production and help your support teams get up to speed.
LONG-TERM SUPPORT
Step 06
Monitoring and updates
We monitor how the product behaves in operation, fix issues and evolve the software as your hardware and usage change.

01/06

[Our team]

An expert team, from consulting to development

We recruit and train engineers, designers and product specialists from the best schools. They combine a real understanding of your industrial constraints with excellent software engineering. The experts who scope your connected product are the same people who code the app and embed the models.

Vincent Defrenet Senior Product Manager

Leads the design of the software around your device, balancing hardware constraints with the way your field users work, all the way to commissioning.

Coline Dorange Lead Product Manager

Defines the software scope of your connected product and coordinates the product, design and development teams to deliver in modules you can test on your hardware.

Clément Dourlens Senior Product Designer

Designs the control and monitoring journeys through workshops with your field users, so screens stay readable in real operating conditions.

Alexie Romeu Senior Product Designer

Runs the design workshops, then designs the app and dashboard interfaces, working closely with the developers to make sure the screens delivered are easy to use.

Surya Ambrose Engineering Manager

Oversees the technical quality of the software chain and structures testing practices, including validation campaigns run on your hardware.

Thomas Hallot Tech Lead

Leads development on the sensitive parts of a connected product: secure pairing, remote updates and protection of the data coming in from the field.

Nicolas Descreux Software Engineer

Develops the app and the communication layer with your devices, over Bluetooth Low Energy, Wi-Fi or cellular networks, with a very hands-on approach to development.

Benjamin Drighès CTO

Advises our clients’ leadership on the architecture of their connected product, including hardware sizing, and upholds our security and scalability standards.

Quentin Massonnat AI Team Lead

Designs the architecture for collecting and storing the history of your field data, and oversees critical development through to go-live.

Félix Monnier AI Team Lead

Trains vision and detection models suited to your device’s data, then optimises them to run directly on the embedded hardware.

Clément de Mareuil Strategy Consultant

Audits how your equipment is used in the field, identifies the use cases that justify software development and scopes the roadmap for your connected product.

[Our work]

Four connected devices run by our software

INDUSTRYBP CASTROL

Securing a filling line with standalone vision AI, on a site cut off from the internet

On a Castrol drum-filling line (Castrol is part of the BP group), protecting operators near the automated zone requires reliable real-time monitoring, yet the site runs entirely offline. We delivered a fully standalone computer vision system: a bespoke model trained on several hours of video and 500 annotated images, specialised hardware selected, then installed on site. The system detects any intrusion in real time and raises the alert before an accident happens, without a single image leaving the plant.

95 %intrusion detection

100 %processing done locally on site

CONNECTED DEVICESYOMY

Embedding AI in a cat feeder that recognises every cat, day and night

YOMY develops a kibble dispenser fitted with a colour and infrared camera, which tailors meals to the needs of each cat in a household. The challenge is identifying each animal reliably, even between two near-identical cats and in the dark. We trained three vision models, including an identification system built for the hardest cases. We then embedded them in the robot after a hardware sizing phase and inference tests in real conditions.

98 %model accuracy

350 msinference time

3models trained

ENERGYOghji

Developing the mobile app that remotely controls the first connected electrical panel

Oghji designs the first digital, connected electrical panel. Its touchscreen lets you configure circuits, track consumption and detect faults across a home or a whole building. We developed the mobile app that extends the panel remotely: it mirrors the control screen and consumption statistics exactly, updated in real time, and alerts the user by notification if anything goes wrong. Pairing a phone requires the user to be physically in front of the panel, so control of the installation stays in their hands.

< 1 secsync latency

INDUSTRIAL ROBOTICSUnista

Configuring pallet layouts from a simple web browser

Unista designs robotic packaging lines, including palletising machines that stack boxes on pallets automatically. Configuring them meant using software installed on clients’ computers, which was hard to use and awkward to roll out. We replaced it with a web application where Unista’s clients build their pallet patterns, either letting the tool arrange the boxes automatically or placing them by hand. They then export a file the machines can use straight away. An admin interface lets Unista’s teams manage access and keep track of the patterns created.

8weeks of development

-50 %support requests when configuring pallets

[Why Galadrim?]

Why clients trust us with their IoT application

Software designed for your operating conditions

We model your device’s constraints and deliver a testable prototype within a few weeks. The product holds up wherever it is installed, including when processing has to happen locally and the network is out of reach.

AI expertise that goes all the way to the device

Our engineers train models suited to your device’s data (images, sensor signals or sound), then size them to run on your hardware with the response times you need.

Engineers who advise, then build

The experts who scope your connected product are the same people who code the app and deploy the platform. With no middlemen, you make decisions faster and the software we deliver matches your real constraints.

Proven expertise

800+ clients supported, 170+ experts and 9 years’ experience building bespoke software, with connected products deployed both in factories and in people’s homes.

Let's bring your project to life together

Contact us

[FAQ]

A few questions our clients often ask