A question is a question until
understanding brings the answer.

Forlais is a British frontier AI research and development company. We investigate new ways of building intelligence, develop original AI architectures and turn our discoveries into technology for people and organisations around the world.

Our work extends beyond AI models, encompassing intelligent systems, software platforms, original computing architectures and the technology that connects them.

We research, design, develop and evaluate new artificial intelligence technologies.

That includes everything from investigating how an AI system learns and reasons to building the models, software, infrastructure and specialised computing technology needed to bring those discoveries into practical use.

EvaEsi is our commercial AI platform, developed from the research undertaken at Forlais.

Intelligence can be represented and developed in more than one way.

Rather than following a single established approach, we investigate alternative methods of computation, learning, reasoning and memory.

Our work begins with original research and continues through engineering, evaluation and deployment. This allows us to develop technologies across the AI system rather than depending exclusively on existing architectures.

Forlais began in 2025 and is incorporated in the United Kingdom as Forlais Group Ltd.

The company was established to undertake original research into artificial intelligence and develop new technologies from that work.

Forlais was founded by Oliver Cooper and Fraser Macdonald.

Friends since their early teenage years, they followed different paths through technology, design, research, aviation, testing and business before bringing their experience together to establish Forlais.

The name Forlais was created by combining elements of its founders’ names, Fraser and Oliver.

A name created from two people who shared an interest in discovering, building and doing things differently.

Forlais is headquartered in the United Kingdom, with its roots in Manchester.

Our research and technology are developed for an international audience, with EvaEsi providing the principal route for our commercial products and services.

Both.

Original research is the foundation of Forlais, but discoveries do not remain confined to research papers.

We develop the models, software, platforms and computing technology that bring our research into practical use. Research and engineering are connected parts of the same organisation.

Our programmes investigate several areas of artificial intelligence, including alternative model architectures, autonomous learning, reasoning, memory, new forms of computation, lossless compression and specialised computing.

Each programme explores its own research questions, with results evaluated through experiments, recorded evidence and technical analysis.

Yes. Original AI model development is central to Forlais.

Our research includes new approaches to computation, memory, learning and the representation of information, rather than relying exclusively on established model architectures.

This research contributes to the original model Forlais develops for EvaEsi.

CNA is an original AI architecture developed by Forlais.

It uses its own form of computation rather than conventional floating-point arithmetic, introducing a fundamentally different approach to how AI systems operate.

Computational efficiency, resource requirements, performance and reliability are central to its development and evaluation.

Most conventional AI models rely heavily on floating-point mathematics.

Floatless AI takes a different approach, using computational methods that operate without conventional floating-point arithmetic.

Forlais investigates these systems as part of its work on alternative AI architectures and more efficient computation.

No. CNA is developed as an original architecture rather than being defined by the conversion of an existing floating-point model into a smaller numerical format.

That is part of how the architecture is constructed, not simply a compression technique applied after conventional model development.

Artificial intelligence has advanced considerably, but that doesn’t mean every important question has been answered or every possible approach explored.

Different computational methods create different possibilities for how systems learn, reason, retain information and operate.

Forlais investigates those possibilities through original research, testing and engineering.

Yes. Lossless compression is an important area of our research.

We’ve investigated how information is represented and stored within AI model weights, including the mathematical and information-theoretic limits associated with their compression.

Lossless compression preserves the original information exactly, allowing it to be reconstructed without alteration.

The Forlais bound describes a result from our research into the lossless compression of neural network weights.

It examines compression limits under specific numerical representations and experimental conditions.

The complete technical meaning depends on those conditions, which are documented alongside the research rather than presented as a universal limit for every AI model.

Yes. Forlais publishes selected research findings, technical explanations and information about its research programmes.

Our public research website provides a place to explore the questions behind our work, the experiments undertaken and the evidence supporting published findings.

Not all of our research is publicly available.

Some material remains under active development, while other work contains proprietary technology, unpublished results or information requiring further verification.

We publish selected findings and research material through Forlais Research, with supporting evidence and appropriate limitations.

Research findings are evaluated through controlled experiments, repeatable testing, technical analysis and examination of the recorded evidence.

We distinguish an experimental observation from a verified finding and do not treat internal testing as independent validation.

The strength of a result depends on the evidence supporting it, not how promising it first appears.

An experiment that doesn’t produce the expected result can still reveal something important.

We investigate what happened, examine the evidence and determine whether the outcome challenges the original hypothesis or reveals a limitation in the method.

Negative results, unexpected behaviour and corrections are all part of scientific progress.

Forlais welcomes relevant enquiries from researchers, universities, institutions and organisations working in related technical fields.

Research collaboration depends on the programme, the nature of the proposed work and the requirements surrounding intellectual property, confidentiality and research integrity.

Our research website provides information about the public programmes and appropriate contact routes.

EvaEsi is the frontier AI platform developed by Forlais.

It brings original research into a connected range of products for individuals, developers and businesses, including conversational AI, software development, research tools and personalised experiences.

Forlais undertakes the underlying research and development. EvaEsi brings that technology into practical use.

Talos is Forlais’ specialised computing programme, extending our work on floatless AI into hardware development.

It focuses on processing technology designed around the requirements of our AI architectures, including local and offline computation.

Talos forms part of our wider work across the AI computing stack.

Talos is Forlais’s own hardware programme for running its AI, rather than a processor sold on its own.

Although connected, each has a different purpose.

Forlais is the research and technology company responsible for the underlying research, development and engineering.

EvaEsi is the commercial frontier AI platform bringing Forlais technology to users, developers and organisations.

Eve is the separate non-profit initiative supported by Forlais, focused on work undertaken for public benefit.

Yes. Forlais brings together people from a range of disciplines, including research, engineering, design, operations and business.

Approved vacancies are published through our Careers page, where you can explore current opportunities and the requirements of individual positions.

Not necessarily.

Developing an AI company involves considerably more than model research. Software engineering, infrastructure, security, design, finance, operations and many other specialisms contribute to the organisation.

The qualifications and experience required depend on the individual role.

No. Forlais Group Ltd is a privately held company incorporated in the United Kingdom.

It is not listed on a public stock exchange.

Forlais maintains a dedicated Investor Relations website for investment-related information and enquiries.

Investment discussions and access to relevant materials are handled through the appropriate channels, subject to applicable requirements.

Forlais publishes public-facing company documents covering matters such as privacy, intellectual property, responsible AI, research integrity and security.

These documents can be found through the Public Documents section of our website.

Whether you have a research enquiry, a business proposal, a question about the company or a request for a meeting, you can get in touch through our official contact channels.

Please include a little information about your enquiry so it can be directed to the appropriate team.