The Edinburgh-based AI firm helps enterprises to build smaller, more focused AI models with greater performance at a fraction of the cost, by taking the output of large models (like ChatGPT) and distilling them into smaller models.

Existing LLMs are overly general and result in very expensive enterprise solutions lacking the required pinpoint accuracy and performance. Malted AI’s technology is the solution for jobs that general AI is not capable of solving. It allows enterprises to apply small language models (SLMs) that solve domain-specific problems with 10-100x cost savings. Instead of doing thousands of tasks moderately well, Malted AI’s SLMs do one task near perfectly.

The firm’s technology is already being used by financial and legal firms in its first pilots, with further roll out planned for the rest of 2024. Malted AI will use the new seed funding to invest in research, product development, and recruitment as it doubles the team size.

“At Malted AI we see smaller as better when it comes to AI. A single, many-billion parameter foundation model is costly, too general and too slow - it’s not fit for purpose for the complex challenges enterprises face,” said Iain Mackie, Malted AI’s CEO and co-founder.

“We are building a world-leading distillation platform that solves the highest value enterprise problems. We do this by using cost-effective small language models (SLMs) that take the most useful parts of LLMs but with a performance exceeding its bigger ‘sibling’ at specific tasks that require domain knowledge. We envision enterprises using Malted AI to build small models to address singular challenges, effectively building a network of proprietary SLMs that are optimised for a single problem.”

Malted AI was born out of concepts that helped founders – Iain Mackie, Carlos Gammel and Federico Rossetto – win the 2022 Amazon Alexa Prize. They beat more than 100 teams to come top of one of the biggest AI competitions.

Today, Malted AI is a 10-strong team with a track record in AI innovation. Its founders include an ex-quant trader, a former-Meta researcher, and a maths champion who lead a team of world-class LLM and applied-AI talent hailing from Deepmind, Amazon, Google, Bloomberg and top PhDs, postdocs and academics from Scottish universities. Malted AI also counts Jeff Dalton, chancellor fellow at Edinburgh’s School of Informatics, as an external chief scientific advisor.

Bryan Gartner, partner at Hoxton Ventures, said: “Malted AI has the capacity to be one of Europe’s most notable AI companies as they flip the thinking and approach to AI adoption within the enterprise.

“Iain, Carlos and Federico are some of the brightest AI minds emerging from the birthplace of European AI – Edinburgh – and have close ties to the University of Edinburgh and the University of Glasgow.

“We are thrilled to be supporting Malted AI as they commercialise their platform and deliver outsized results for businesses in the tiny package of small language models.”

Malted AI has raised £7m in total investment, thanks to a £1m pre-seed round led by Creator Fund in August 2023.