
The Right Tool For Every Step Of Your AI Journey
Our own platform, proven in critical operations, plus a careful selection of products from other companies that we implement for you. All chosen for one thing: AI you control, host where you want, and can leave without strings attached.

Four Products.
One Way Of Working.
Olympe is our own platform. Constelio, skilder.ai and Hermes Agent are built by other companies: we select them, implement them, connect them to the systems you already run and keep them under your control.
Olympe
Wrap your code and systems into reusable, governed components. Runs critical operations at global leaders.
Discover Olympe → We implement · open sourceConstelio
A runtime for long-lived agents that each own an operational duty, and learn to do it in code.
Discover Constelio → We implementskilder.ai
Your teams' know-how, as business skills inside Claude, ChatGPT and Copilot.
Discover skilder.ai → We implement · open sourceHermes Agent
A self-improving agent that remembers, learns skills on the job and runs on your own infrastructure.
Discover Hermes →Olympe:
The Code Orchestrator
Olympe wraps code into reusable components, so you stop reinventing the wheel and focus on what matters. It is the platform global leaders run procurement, supply chain and finance on, and the place where what AI produces becomes tested, versioned and governed.
Why teams choose Olympe
- Modular development. Turn your codebase into easily discoverable, reusable components.
- Data-driven. Build data flows that get the job done with the Logic Editor.
- AI-enabler. Connect enterprise systems, business processes and AI agents, from exploration to production.
- Code as data. Powered by Neo4j graph technology: you always know what is where.
- Common knowledge. Connect your IDE, sync your Git and code in your favourite language.
- Distributed. Deploy anywhere, from cloud to edge, with the Olympe Runtime.
Olympe Features
Codebase Integration
- Two-way Git syncLeverage your existing code and legacy systems, without duplicating them.
- IDE integrationKeep your favourite development environment and settings.
- Reusable capabilitiesTurn code into standard components in seconds, with AI-generated signatures.
- BranchingMany developers on one project, at scale.
Capability Catalogue
- Open-source catalogueReuse community libraries to go faster.
- Private librariesYour own capabilities, with fine-grained roles and permissions.
- InceptionInspired by the C4 model: drill down or promote components into business objects.
- Semantic searchFind capabilities by asking in plain language.
Visual Composer
- Front to back in a secondExpose any function as a web service.
- Data-flow editorCompose software around the output you expect.
- AI-assistedExplore and use component libraries with AI search.
- Data modellingA federated model across all your data sources.
Full Governance
- Embedded documentationGenerated by AI for each capability, always up to date.
- List usageSee in one click where each capability runs.
- StandardisationInput, transformation, output: the context AI needs to build reliably.
- Lifecycle managementA status for each component, to keep technical debt down.
“Olympe platform streamlines the software development process and links it directly to business needs, empowering CIOs to extract more business value from their IT assets.”
Constelio:
Agents That Own A Duty
Most agents run a task and stop. Constelio agents take on an ongoing duty, such as "keep supplier data consistent across our ERPs", and keep at it for months. Day one, AI does most of the work. Over time, what keeps working becomes reviewed skills, then plain code. AI stays for what's new and unusual, so costs go down instead of up.
How it works
- One agent, one mission. Each agent perceives, decides, acts and reflects on its duty, for as long as the duty exists.
- A learning loop. Successful patterns are distilled into skills, then tools, with a human approving every promotion.
- Secure by default. Every runtime is identified and approved before it can do anything. Every action is signed and logged.
- Budgets and model choice. Per-mission rules decide which AI providers may be used and how much they may cost per day.
- Runs on your servers. A simple Linux install, with Anthropic, OpenAI or fully local models such as Ollama or vLLM.
skilder.ai:
Your Know-How In Every AI Assistant
Your people already use AI assistants, with or without permission. skilder.ai turns your teams' methods into business skills that run inside the assistants they use every day, under your rules. No more "shadow AI", and no more reinventing the same prompt in every team.
What you get
- Skills, not prompts. Each skill packages a method, the tools it uses and the expected output. It loads only when a request needs it.
- Your permissions, nothing more. A skill runs with the user's own rights and can't open anything they couldn't.
- Fully traceable. Every run is logged per user, with versioning, rollback and an exportable audit log.
- Enterprise-ready. SSO, role-based access, isolated workspaces and connectors from your own registry.
- Owned by the experts. The team that publishes a skill maintains it, and everyone gets the update.
Hermes Agent:
An Agent That Learns On The Job
Hermes Agent is an open-source agent built by Nous Research. It remembers across conversations, turns experience into reusable skills and works with any model, including local ones. We deploy it for you, connect it to your systems and put it under the same governance as everything else we build.
What Hermes brings
- Memory. It keeps what it learns across sessions and can search past conversations.
- Skills from experience. It creates skills from what worked, and refines them as it uses them.
- Automations. Scheduled tasks and a gateway to the messaging tools your teams use.
- No lock-in. Open source, any AI provider or a local model, from a small server to a GPU cluster.
No Lock-In.
No Strings Attached.
We recommend what's right for you, not what's easiest for us to sell. Wherever we can, we choose open source and models you can run yourself, so your business secrets stay with you and no AI vendor sets your prices.
Stays With You
Hosting in Switzerland or on your own servers, under Swiss and European rules.
Your Choice
Anthropic, OpenAI, Mistral or a local model. Switch when it makes sense.
Yours To Keep
Open formats and open source, so what we build together stays yours.
At Every Step
Approvals, audit trails and budgets, whatever the product.
Not Sure Which
Fits Your Business?
That's our job. In one conversation, we'll map your needs to the right tools, and tell you where you don't need AI at all.
