Alinia AI raises $2.4M for safe deployment of generative AI

Alinia AI, an alignment platform that enables companies to safely and controlled deploy generative AI, guided by their policies and business, has raised $2.4 million in Pre-Seed funding. 

Significant adoption of generative AI is underway, but its rapid adoption comes with substantial challenges. 

An example is managing control and safety in the face of risks to prevent inappropriate or brand-damaging content.

There's also the need to navigate upcoming regulatory efforts like the EU AI Act, Biden's Executive Order, or Canada's Bill C-27, which demand rigorous safety checks and clear documentation.

Evaluating the performance and reliability of generative AI applications from both safety and business dimensions is a top obstacle for the enterprise today.

This is especially challenging for companies when leveraging advanced techniques like fine-tuning and RAG, maintaining LLM versions, and coordinating across teams including machine learning, governance, and business domain experts.
 Ariadna and Carlos Muñoz Ferrandis co-founded Alinia AI. 

Font Llitjós was the former Head of ML Platform at Twitter, where she established responsible AI at the company level. Before that, she was Director of Development at IBM Watson in the early days of enterprise AI, where she led teams of software engineers, designers, and researchers to build the first generation of AI assistants and other AI enterprise solutions. She has a PhD in NLP from Carnegie Mellon University. 

Carlos is the former Tech & Regulatory Affairs Counsel at Hugging Face, where he co-led governance of open LLMs and was heavily focused on the EU AI Act. An open-source and standards PhD graduate of the Max Planck Institute for Innovation and Competition in Munich, he was also an external advisor for OECD AI on AI regulation.

"The technology is not production-ready for business scenarios. Enterprise leaders are rightfully afraid of making headlines because their generative AI applications go rogue. Reputational harm keeps them up at night," says Ariadna Font Llitjós, co-founder and CEO of Alinia. 

"Using a process known as Alignment, we ensure AI systems adhere to expected behaviours, policies, and regulatory requirements – helping enterprises seize the promise of this new technology while minimising the risks and potential harm."

Font Llitjós saw firsthand, at Twitter, the impact the unintended bias of training data and ML algorithms can have on vulnerable populations. 
She brings this learning to Alinia so that other companies can increase their chances of getting generative AI right for their employees and customers.

The Alinia Alignment Platform aims to provide an integral AI governance approach – which currently requires multiple tools – ranging from evaluation and real-time monitoring to cutting-edge optimisation techniques, coupled with documentation features facilitating compliance at every stage of the LLM lifecycle. 

It enables customers to have clear and accurate evidence of how LLM-powered applications behave in different enterprise scenarios for specific tasks and abiding by specific rules.

"In my experience at Hugging Face, governance and safety aspects of LLM development require incredible time and effort. But when you are about to release a foundation model that will be used millions of times, governance and safety are non-negotiables," says Carlos Muñoz Ferrandis, co-founder & COO.

Speedinvest and Precursor led the round with participation from KFund and notable angels including Clem Délangue & Thom Wolf (Hugging Face Co-Founders), Tom Preston (GitHub Co-Founder), Xavier Amatriain (Google Core AI VP), and Oriol Vinyals (Google DeepMind VP of Research).

Fred Hagenauer, Partner with Speedinvest, says: 

"Ari and Carlos have some of the best profiles we've seen to tackle the alignment problem in the generative AI space. Their experience and knowledge of the early days of enterprise AI, RAI, and the LLM governance required today is something truly unique. For us, it was a no-brainer to be the first ones to support them."

Oriol Vinyals (VP Research, Google DeepMind): 

"The safe development and use of powerful machine learning models should be of concern to all, and diversity is key in this field. That is why it is so important that Ari and Carlos are addressing this highly nuanced challenge with Alinia."

Tom Prestner, Co-Founder, GitHub, says: 

"Alinia comes at the right moment in time. The majority of companies are still in the experimentation phase, playing with LLMs internally for non-critical business ops. This will change soon."

Alinia will use the funding to hire additional talent to build the Alinia Alignment Platform to create an end-to-end alignment process focused on safety and regulatory requirements – enabling safe and inclusive use of generative AI across LLM modalities and languages.

Lead image: Alinia Ai. Photo: uncredited.