Loreley A.I.

Your A.I., on your hardware, with your data.

A proof of Concept

How it started ?

The idea of a fully on-premises A.I. for the hospitality industry grew naturally while observing the issues generated by cloud-based A.I. solutions:

Users' data being sent to US and China
Network latency from remote location
No fine control over model behaviour
Closed source models deactivation
Prices surges

The first A.I. meta-model was developed and presented in two weeks – with voice support already built-in – running on a 2’500.00 CHF personal computer.

A first Customer

Who trusted the project ?

The technology was first pitched in June 2025 to Explora Journeys (MSC Group) as a fully local, safe and multi-cultural onboard A.I. solution for their fleet of Luxury cruises Ship.

Over the course a few months, the meta-model evolved to a point it was integrated into all core aspects of the Ships, and capable to process up to 250 requests in real-time (< 3 seconds latency) with a hardware of 75’000.00 CHF

In April 2026, Tanaquil – the meta-model name in Explora Journeys – became operational on Explora I.

A new Product

Where is the project used ?

Tanaquil is now being deployed on the whole fleet of Explora Journeys – 4 Ships – and continue its development as a “virtual concierge”, always available to help the Ship’s Guests and Hosts:

Full support more than 14 languages – written and spoken
More than 40 languages in a short term
Abilities to book restaurants, SPA treatments and other services
Aware of the Ship information (position, weather, security information etc.)
Gently promoting Explora Journeys in-cruises offers (excursions, products, restaurants, future cruises, etc.)

A bright Future

What are the next perspectives ?

It is time to go one step further and evolve Tanaquil into an even more efficient solution.

A new project, named Loreley, is benefiting from all the invitation brought by Tanaquil – plus a new unique technological core, allowing for vision, natural conversation, 3D avatars rendering via an artificial nervous system.

Why use our A.I. ?

Meta-model | Infrastructure | Framework | Integrated

Meta-model

Continuously up-to-date

Loreley A.I. is an aggregate of multiple A.I. models – not only LLM, but a mix a various specialized systems.

We add and update models as needed in a matter of days, to ensure we fit your needs and resources limitation.

From the developer's perspective, Loreley A.I. remains a standard LLM, with standard APIs – one can use the same tools, documentation and knowledge accumulated with other technologies.

Infrastructure

No cloud & no third-party

Loreley A.I. is meant to run on-premises, i.e. on your own hardware. This guarantee total control and sovereignty over your data – nothing never leave your environment.

We design, build and operate intensive computational clusters – CPU, GPU, FPGA – fit to run and train A.I. models.

Loreley A.I. infrastructure is scalable – adding a compute node takes a few minutes after physical installation.

Although we are local-first, we do support running Loreley in a Cloud infrastructure if required – this however remove the data locality guarantee.

Framework

What makes Loreley A.I. different at the core ?

We developed a complete framework to make Loreley adaptable and efficient in your ecosystem; We introduce two innovative technologies:

An events hub, connecting A.I. models and tools to numerous data sources and orchestrating information flow
A domain-specific language for designing Loreley A.I. nervous system connections

We do also keep an upstream compatibility with existent state of the art technologies and practices – MCP, A2A, agentic workflows, RAG, fine tunning, memorization, etc.

Integrated

How Loreley integrates your organization ?

We proceed with rationalism and prioritize truth – here’s our roadmap:

Use cases definition: what are you trying to achieve ? What are your problems ? How can A.I. helps ? Is another solution more fit ?
Available resources: What are your actual needs & capabilities ? Can you host a compute system ?
Proposal: Milestones to achieve a successful project – and how it will evolve over time.

Why on-premises ?

Closed loop

Isolated infrastructure with complete control over data input and output.

Integrated with Loreley platforms
Identity
Authentication
Authorization
Logging
Metrics
Reporting
Alerting
Etc.
Inference pipeline – Respond to/in text, audio and picture
RAG pipeline – Consume from any data sources

Edge environments

Runs next to your data & activities.

Remote location – Cruises ships, oil rig, industrial zone, etc.
Secure sites – Data and processes remains local, within a completely isolated environment
Zero network latency – Surveillance systems, trading, etc.

Compatibility

No cost associated with token usage.

API – Fully compliant with OpenAI’s API means drop-in replacement for cloud-based services
Integration – Fully compliant with MCP and support for MCP gateway
GUI – End-user interfaces with speech and complex data support

Data sovereignty

Conversation, documents, code, PII, … remains local..

Personal information – PII (users’ data customers’ information, etc.) remains local
Intellectual property – Your code, documents and sensitive data are processed locally

Your Title Goes Here

Unlimited tokens