1. Home
  2. Platform
The NatusAI platform

One sovereign AI platform for your data, AI assistants and applications

NatusAI connects documents, systems and history into one picture of how your business works. AI assistants answer from it with sources, people approve what matters, and everything runs where you decide.

Available NatusAI Lyra Planned v1 NatusAI Cygnus Roadmap five more products

Glowing glass cubes holding connected node graphs, linked in space: one connected platform
What your business can do

AI that works with the knowledge of your business

Bring your information, people and decisions into one connected context. Give your teams practical help they can understand and trust.

Explore our capabilities

Bring your information together

Connect documents, messages and business systems so useful information can work together.

Give AI the context of your business

Connect customers, projects, products and decisions in a shared business memory.

Find answers with evidence

Ask questions in everyday language and follow the supporting information.

Coordinate the next step

Connect knowledge to AI-assisted tasks, applications and authorised tools, with people in control.

Understand decisions and change

Review the information, rules and recorded actions behind an outcome, and see what changed.

Keep ownership and control

Define application ownership, hosting and approved AI connections from the start.

Proof, not promises

Proven on the same engine as our client projects

Our team delivered these results for clients in regulated industries, on the engine behind the platform.

3 days → <1 s

complex relationship queries

Large European bank

92%

faster response to customer enquiries

European telecom operator

0

expensive hardware upgrades for a country-wide grid twin

Electricity distribution operator

Results reported by each client at project close. NatusAI Cygnus, NatusAI Auriga, NatusAI Aquila and NatusAI Lyra are how we are turning these capabilities into products.

Honest status

What you can use today, and what is on the roadmap

We separate what is proven from what is coming. Both are labelled, and nothing is presented as available before it is.

What you can use today

  • A delivery engagement led by our senior team, built on the engine proven in the projects above.
  • NatusAI Lyra, our interface library for dashboards, maps, documents and conversations. Available
  • Your applications, owned by you, running in your environment, with the exit documented from day one.

What’s on the roadmap

  • NatusAI Cygnus first release: the connected data foundation, released only after its defined acceptance tests pass (v1 scope · Cygnus roadmap). Planned v1
  • NatusAI Orion, NatusAI Auriga, NatusAI Aquila, NatusAI Draco and NatusAI Gemini: self-service application building, AI coordination, document intelligence, operations and verified recovery. Roadmap
  • Dates under NDA. We share the roadmap in a briefing.
A delivery team and business users co-designing an application at a whiteboard (illustrative image)
Today, every engagement starts with our team and yours in one room.
Architecture

Three layers, seven connected products

Each product has one clear job. Together they turn connected data into applications your people can trust.

Compare all seven products
Layer 01 · Connected foundation

A connected foundation you own

Relationships, history, location and search in one place, so AI works with the full context of your business.

The application lifecycleRoadmap

From a business request to a running application

The path our team follows in every engagement today, and the path NatusAI Orion is designed to make self-service. See the intended Orion Workspace.

An observatory-like glowing sphere above floating dashboard panels: the intended Orion Workspace (illustrative render)
  1. Describe

    State the outcome, the people involved and the rules that apply.

  2. Create

    AI helps draft the application; your team reviews something concrete, not a specification.

  3. Verify and release

    Checks run against agreed acceptance criteria. A person approves the release; AI never approves its own work.

  4. Operate and improve

    Monitoring shows how the application behaves, and every improvement links back to the original request.

Improve once. Choose what to share.

Proven improvements can be reused across applications, while each organisation keeps its own data boundary. Nothing is shared without your decision.

Control by design

How the platform de-risks AI

Four controls run through every layer. They are what your risk, security and audit teams will ask about first.

  1. Ownership

    Your data, the models you choose and the applications we build stay yours, in your environment. The exit is planned at the start.

  2. Evidence

    Answers cite the documents and records they come from, and history shows what was known at the time.

  3. Human review

    AI proposes, people decide. Consequential steps wait for a named approver, and no agent can grant itself more access.

  4. Recovery

    Backups only count when a restore has been tested. The platform keeps a route back if a component or AI provider fails.

How we handle sovereignty, governance and security

A specialist checking an answer against its linked source documents before approving it (illustrative image)
AI proposes; a named person approves anything consequential.
Night view of a coastline with lit ports, a container ship and light routes: a connected worldwide twin
One shared picture

A worldwide twin starts with connected facts

A ship, its cargo, its route, the weather and the reports about them are facts about one situation. Connected, they answer questions like: “Which cargo was on vessels inside the storm area yesterday, and which reports explain the risk?”

The same approach works for a power grid, a bank’s customer relationships or a hospital’s patient flow. It is how a national grid twin and a bank’s investigation workflow were delivered.

Illustrative example. Worldwide twins are a demanding first domain for the planned NatusAI Cygnus release; this is not a live global service.

Glowing server towers on a cliff linked by an arc of light to a second island: operations with a route back (illustrative render)
Continuity by designRoadmap

Observe, investigate, verify, and recover deliberately

Self-monitoring and self-healing describe the target experience. Their scope is governed by declared permissions, resource limits, acceptance checks and independently controlled recovery.

  1. Know what changed

    Connect operational findings to the affected application, release and environment. Compare desired configuration with observed behaviour.

  2. Repair within policy

    Registered, preauthorised recovery actions can have bounded operating rules. Work outside those rules must stop or escalate; an agent cannot approve its own authority expansion.

  3. Prove the way back

    The intended role of NatusAI Gemini includes preserved recovery points, restoration and verification. A recorded backup is not the same as a demonstrated restore.

Phoenix remains a proposed name for automated repair and incident recovery. It is not a selected additional product. Recovery times and service commitments require a qualified deployment.

Explore NatusAI Draco Explore NatusAI Gemini

Why this direction

Better tools make a more connected approach worth pursuing

AI-assisted development changes how a team can work with typed contracts and generated clients. Agent-facing interfaces make capabilities easier to discover. Dense storage makes bounded, larger-than-memory workloads practical design targets. And operational teams need a history they can trace and reproduce.

These are the reasons for the architecture direction. Benchmarks, compatibility matrices and release acceptance still have to provide the evidence.

Built with AI, designed for AI: the seven principles

Glowing paths passing through checkpoint gates to a crystal tower: AI work checked before it is accepted (illustrative render)
Questions

Questions about the platform

Short answers for decision-makers, with the detail for your technical team below.

For your technical team
  • One unified engine: one identity, one commit boundary, one authority model and versioned contracts across data, logic and services.
  • Discoverable, not privileged: HTTP/RPC, OpenAPI, JSON Schema, MCP descriptors and generated TypeScript clients expose the same typed services to applications and approved AI agents.
  • A verified library loop: discover, develop and qualify, then approve and reuse. AI workflows are used to create and maintain parts of the ecosystem; no measured token-saving or cost-reduction claim is made.
  • Planned Cygnus v1 scope: embedded, development and single-server production profiles; clustering later. Release scope and acceptance gates; what comes after v1.

Is the NatusAI platform available today?

Parts of it. NatusAI Lyra is available, and our team delivers applications on the engine proven in our client projects. NatusAI Cygnus is planned for its first release, and the other five products are on the roadmap. We label each one clearly and share dates under NDA.

Do we have to buy seven products?

No. The seven names describe clear responsibilities inside one platform, not seven subscriptions. Commercial terms are agreed per engagement; ask us for a scoped proposal.

Can the platform run inside our own data centre?

Yes. It is designed for on-premises, European sovereign cloud or hybrid deployment. The planned first release of NatusAI Cygnus runs on a single server.

Which AI models can the NatusAI platform use?

The ones your policy allows: hosted, EU-hosted or running locally. Model choice is recorded in the architecture.

What happens if an AI provider becomes unavailable?

The platform design keeps an independent route to restore and operate without it. It is a design requirement, not yet a service guarantee.

How would AI usage be funded?

The commercial model remains open. Options include platform-funded APIs with metered usage, an included allowance, or customer-provided credentials. No provider, price, unlimited-use allowance or reseller arrangement has been selected.

What can we discuss today?

A concrete business problem, the information involved, the people who own it and the deployment constraints. We can separate an engagement around current tools from dependencies on planned platform capabilities.

Next step

Start with one business question

Tell us where work is slow, risky or invisible. We will show you what the platform can do for it today, and what comes next.

Or write to contact@natusai.com

A product owner, a security officer and an analyst reviewing an internal app before release (illustrative image)