[Innovative AI Experiences]

Reinvent your product's UX with Generative AI

Galadrim designs and develops innovative AI products that transform the digital experience for your customers and users, inventing the uses of tomorrow.
Proactive and voice interfaces, connected to your product's real functions
+30 experiences already deployed, with measurable results
+140 senior tech profiles from Polytechnique, Centrale or ENS
Chat with an expertDiscover 5 AI experiences
MathexIA - AI-generated maths training.
+800 organisations entrust us with their tech and AI projects
[Our Expertise]

The innovations your customers will demand tomorrow

Voice command, brand companion, screens generated on demand or anticipation of needs: we design the form that suits your use, and connect it to the real functions of your application.
Dynamic interfaces

Dynamic interfaces

Artificial intelligence rethinks web and mobile navigation by generating tailor-made interfaces on the fly. Your applications adapt instantly to the context and requests of each user, offering total personalisation and fluid journeys without complex menus.
Brand companions

Brand companions

We design specialised virtual assistants and immersive avatars that embody your identity. Beyond simple text chatbots, these companions support your customers with empathy, pedagogy, and expert knowledge of your products at every stage of their journey.
Voice experiences

Voice experiences

Free your users from traditional graphical interfaces through “Voice-First” approaches. We integrate natural and conversational voice commands that allow you to control your applications, interact with virtual coaches, or navigate in a totally intuitive way.
Augmented Co-creation

Augmented Co-creation

Transform your users from simple consumers into creators. We integrate AI directly into the heart of your workspaces (canvases, editors, product configurators) so that it acts as a real-time creative partner, generating suggestions, visuals, or plans as the user shapes their idea.
Predictive Navigation

Predictive Navigation

Transition from a purely reactive interface to a proactive experience. By analysing real-time context, our AI models anticipate your users' needs and suggest the exact action or information even before they request it, minimising friction.
Simulations and World Models

Simulations and World Models

Prepare for tomorrow's experiences with World Models capable of understanding physics and spatial dynamics. We support you in prototyping interactive virtual environments, immersive try-ons, and visual simulations of unprecedented realism.
Two people from behind consulting a laptop displaying an AI chat interface
New AI models open up UX possibilities that will soon render your traditional interfaces obsolete. Voice interaction, dynamic display adaptation, real-time personalisation: today, we have the building blocks to create digital products that radically transform your users' experience.
Benjamin DrighèsAI CTO
[Our method]

Our method for designing, testing, and deploying your AI assistant

  • Framing
    Step 01

    Journey to transform and expected service level

    We start from the precise moment AI should intervene in your product, then we define with you what makes the experience successful: perceived response time, scope of covered requests, and expected behaviour in case of doubt. These thresholds determine production deployment, set before the first development.
  • design
    Step 02

    Scope of actions and safeguards

    We list with you the actions the assistant actually triggers, those it proposes for confirmation, and those it refuses. This scope is maintained by design: a request outside it is rejected at the input, a response outside it is blocked before reaching the user.
  • Prototyping
    Step 03

    Choice of models tested on your real case

    We compare the components of the chain – speech recognition, language model, and speech synthesis – using your own content and your own users. This test bench determines the model chosen for each, and which ones we host ourselves when response time requires it.
  • Intégration
    Step 04

    Assistant connected to your business APIs

    The same team designs the experience and develops the application: we connect the model to your product's functions and your existing systems, securely and robustly with your information system.
  • Evaluation
    Step 05

    Reference dataset and adversarial tests

    We build a set of reference cases with your business experts and measure each version against it, then we test the assistant on out-of-scope requests and bypass attempts. You receive a quantified result before opening it to users.
  • Sustained operation
    Step 06

    Monitoring real conversations and improvement

    We instrument in-service exchanges, including latency, cost, and satisfaction, and feed failed cases back into the reference set. Each iteration is measured against the previous version.
01/06
[Why Galadrim?]

Why do companies entrust us with their AI experiences?

Proven expertise

800+ clients supported, 170+ experts, and 9 years of experience in AI product development, with voice and conversational assistants in service in health, hospitality, and training.

Assistants that act within your systems

An assistant that merely describes your features remains a demonstration. Our engineers connect the model to the real functions of your product – search and booking, payment, controlling connected equipment, or writing to your back-office – with each action remaining subject to the rules you set.

Sourced answers and explicit safeguards

Our assistants respond based on your own documentation and cite the original reference. A response without a source is blocked before the user, personal data is anonymised before any call to the model, and the authorised scope is checked upon entry and exit.

An innovation that belongs to you

Code, prompts, evaluation datasets, and intellectual property revert to you with each delivery, including inventions arising from the project. Our engineers have previously been declared inventors on a patent filed by a client at the end of a mission.
[Projects]

Five AI experiences in operation for our clients

TRAINING IN FINANCETraining You

Conducting an M&A interview with a vocal interviewer who adapts their questions

Training You prepares students for interviews with investment banks and consulting firms, an oral exercise that is difficult to rehearse alone. Galadrim developed the voice training application integrated into its course platform: the AI interviewer asks their question aloud, the candidate responds orally, a Large Language Model scores the answer and prompts them on weak points. An internal ranking adjusts the difficulty throughout the interview, and the candidate receives an audio debriefing question by question.
5 000questions in the database
30 minper interview
+8 000students trained
Training You Logo
INNOVATION IN HEALTHCAREGenolier Innovation Hub

Translate a live conference into each participant's headphones

The Genolier Innovation Hub brings together researchers, doctors, and industry professionals from across Europe in its 300-seat auditorium, where each session is held before a multilingual audience. Galadrim developed MyGIH, the hub's mobile application, and integrated the simultaneous translation chain: the sound from the control room feeds an AI agent that transcribes the speaker, translates their words by segmenting the text at breaks in meaning to preserve context, then re-broadcasts the translated voice and its transcription into the room. Each participant chooses their language and listens through their own headphones.
300auditorium seats
4languages
1patent filed
Genolier Innovation Hub Logo
HOSPITALITYLeev

Voice-control hotel booking, doors, and air conditioning

Leev operates a hotel-apartment complex in Paris where the entire stay can be controlled via the mobile application developed by Galadrim. We integrated a voice assistant that executes journey actions: the user makes a request orally, a language model interprets it and directly calls product functions such as the booking engine, smart lock, air conditioning, and payment. Each component of the voice chain was compared with its alternatives to maintain response time, and payment is triggered after explicit user confirmation.
2 à 3 sbetween command and action
4voice-controllable actions
Leev Logo
HEALTHCARE INDUSTRYUrgo Medical

Guiding a nurse in choosing a care protocol, with supporting sources

Urgo Medical equips 160,000 healthcare professionals with Healico, its wound tracking application, and wanted to make the expertise of its specialists available to non-specialised nurses. Galadrim developed the conversational assistant backed by a closed corpus of validated medical references: it queries the indexed literature, cross-references the pathology and the state of the wound bed to propose a protocol, and systematically cites the original scientific source. Safeguards rule out any response lacking a source or related to diagnosis, which keeps the product outside the scope of a medical device.
160 000healthcare professionals equipped
12pathologies covered
7languages
Urgo Medical Logo
PREPARATORY CLASSMathexIA

Composing a personalised mathematics training for each prep school student

Commercia prep school wanted to accelerate the progress of its ECG stream students in mathematics, a subject that is self-studied and where teacher feedback often arrives belatedly. Galadrim developed MathexIA, an application that assesses each student's level, generates on-demand tutorials and quizzes tailored to their weaknesses, and then marks their work. Each student thus progresses along a personalised path, and the teacher opens their session by addressing the actual sticking points of their class.
+3 000student users
+5 000exercises generated
MathexIA Logo
They talk about us
[Our team]

A team that advises you before building

We recruit and train passionate engineers and experts, from top universities, who combine expertise in artificial intelligence and software engineering skills. This hybrid positioning allows us to design your bespoke AI experiences and industrialise them for production.
Benjamin Drighès
Benjamin Drighès CTO Data & AI
Advises our clients on their product strategy, their long-term AI transformation, and their structuring choices of technologies and approaches, and defines the quality standards for our AI teams.
Quentin Massonnat
Quentin Massonnat AI Tech Lead
Designs the architecture of conversational chains, including speech recognition, model orchestration and speech synthesis, and oversees developments until production deployment.
Marc-César Garcia-Grenet
Marc-César Garcia-Grenet AI Engineer
Develops complete applications in production and agentic systems based on Large Language Models, and advises clients on their product on a daily basis.
Félix Monnier
Félix Monnier AI Engineer
Builds the document search engines that power our assistants, and measures the quality of their responses against validated test cases from your business experts.
Lucien Maillard
Lucien Maillard Principal AI Strategy Consultant
Frames AI experience projects with client executive boards, identifies the journey to be transformed and quantifies the achievable gain before any development.
Pierre-Antoine Dornic
Pierre-Antoine Dornic Principal AI Strategy Consultant
Maps AI use cases, prioritises them by ROI and feasibility, frames the executive boards' roadmaps and manages their deployment until team adoption.
Eva-Garance Tison
Eva-Garance Tison Lead AI Product Manager
Transforms business needs into adopted AI products: frames projects with clients, prioritises the roadmap, manages team Delivery and ensures the impact of delivered solutions.
Naïs Schietecatte
Naïs Schietecatte Senior AI Product Manager
Manages the product cycle of AI solutions, from functional specifications to the control interfaces used daily, in close collaboration with engineers and business users.

Let's bring your project to life together

Contact Us
[FAQ]

Some frequently asked questions from our clients

It acts. We connect the model to your application's real functions, such as search and booking, payment, control of connected equipment, or writing to your back-office, with each action remaining subject to the rules you set.
We back it with your own documentation and require a source for every response generated. A response lacking a validated reference is blocked before reaching the user, and the authorised scope is controlled upon entry and exit.
Working with your experts, we build a set of reference cases and score each version against them, then we test the assistant on out-of-scope requests and attempts to bypass. You have a quantifiable result before commissioning.
The duration depends on the scope of actions to cover, the number of systems to connect, and the level of guarantee expected for the responses. We provide this framework after an initial discussion about your context, and each stage concludes with a verifiable result.
No. Our assistants operate in multiple languages, including simultaneous translation of a live broadcast speech directly into each listener's headphones.
In France or the European Union, with a host defined with you according to the sensitivity of this data, or directly on your own infrastructure. Personal data is anonymised before any call to an external model.
Yours to keep. Code, prompts, evaluation sets and documentation belong to you with each delivery. The transfer of rights is contractual from the start, including for patentable innovations resulting from the project.
Each project is sized according to your scope and priorities. Describe your context to us, and we will get back to you with an initial assessment.
[Contact us]

Let's bring your project to life together

We work with all types of clients, across all sectors. Whether you're an entrepreneur or managing a large organisation, a tailored team will meet your needs.

Over 800 companies have trusted us to create their web, mobile, and AI products

Your enquiry

We'll get back to you within 1 hour.