Malted AI

Dr Sandy Nairn recently published "AI Comes of Age on Language," where he unpacks the benefits and limitations of deploying AI with our co-founders Iain Mackie (CEO), Carlos Gemmell (CTO) and Federico Rossetto (Chief of Engineering). Sandy has a wealth of experience understanding long-term technology trends, having previously been Chairman of Franklin Templeton Equities Group, Chief Investment Officer at Scottish Widows, and written multiple best-selling books, including "Engines that Move Markets". A couple of key takeaways: 🧠 Domain knowledge is essential: In many applications, technology should not be agnostic. It requires specific domain knowledge if it is to be effectively applied. Thus, a partnership between enterprises and technical AI experts is critical to fill these two-way knowledge gaps. 📊 Data is critical: The ability to access, collate and process unstructured data will be invaluable, and the battle over property rights is only just beginning. Enterprises who can harness their data will be best placed to leverage AI and future-proof their businesses. 🔍 Traceability essential: Many Large Language Models (LLMs) (like OpenAI) are not naturally deterministic. In other words, asking the same question can result in different answers. While Small Language Models (SLMs) can be version-controlled and determinism enforced for sensitive use cases. These types of technical considerations will be essential to safely deploy AI. Well worth a read: https://lnkd.in/eiwy9ifF

Malted AI has turned one! 🥳 This past week, we celebrated this milestone in beautiful Loch Lomond. We had an incredible time reflecting on our achievements from the past year and creating an environment for our fantastic team to connect further and energise for what's to come. In the last year, we: • Raised two rounds of investment • Proved that a pipeline of SLMs can solve problems where general AI fails • Grew to a team of 20, attracting top talent from startup unicorns and the best research labs in Europe • Developed leading research in knowledge distillation and the optimisation of SLMs This week wasn't just a celebration — it was also a great opportunity to strengthen bonds within our team and ensure our vision for Malted AI remains aligned. 🌾🏴󠁧󠁢󠁳󠁣󠁴󠁿

It was great to speak at FinTech Scotland events on innovation (Edinburgh Futures Institute) and on AI regulatory use cases (Barclays)👇

We had an exciting week presenting our unique approach to AI at two fantastic events, where we discussed regulation and financial innovation. (1) We presented Malted AI's approach to Small Language Models (SLMs) at the Fintech Scotland Strategy Partners assembly at the Edinburgh Futures Institute. (2) At the Financial Regulation Innovation Day at Barclays Eagle Labs, we showcased how our platform tackles regulatory use cases. Our CEO, Iain Mackie, presented how SLMs are a game-changing technology for regulated industries that need accuracy, trust, and scale. He also touched on the limitations of applying closed-source LLMs to these critical use cases. We see the future being driven by SLMs who are experts at these specific tasks. Thank you to FinTech Scotland for the invitation to share our thoughts and be part of this fantastic community.

It was great to attend and speak at two fantastic events in Scotland last week: 1️⃣ BlackRock's Techfest: Joined a panel with Tony Ashraf (BlackRock), Nicola Anderson (FinTech Scotland), Nick Jones (Zumo), Pardeep Cassells (AccessFintech), and Christina M. (Space Scotland) to discuss Scotland's innovation across sectors like AI, fintech, space, and digital assets. 2️⃣ CIO Scotland: Engaged with leading CIOs from industries including finance, healthcare, government, and oil & gas, discussing the key challenges and opportunities. Key takeaways: 📊 Data is King: As Kshitij Kumar (Haleon) put it best, "data is king". Enterprises must focus on accessing and leveraging data securely and at scale. However, this is a significant challenge for most enterprises. 😀 Outcomes Over Technology: Whether it's NHS 24 improving patient triaging or BlackRock supporting investment decision-making, the focus must be real-world use cases that benefit staff and customers. Collaboration across AI, data, products, and compliance is essential for positive outcomes. 🏴󠁧󠁢󠁳󠁣󠁴󠁿 Scotland: Scotland is thriving with innovation across multiple industies. Despite the variety, the one thing every business had in common was leveraging world-class local talent. At Malted AI, we partner with enterprises to build accurate, trustworthy and secure AI solutions - leveraging distillation to get the most value out of your data.

Enterprises are facing significant challenges in extracting value from AI. They require factually accurate, flexible, and secure solutions. Moreover, enterprises need internal expertise to work with a complex technology that the world is still researching. We have had numerous conversations with different organisations where Proof of Concepts are struggling to show ROI. We believe that a significant part of the problem is general-purpose Large Language Models (LLMs) being applied to tasks that require domain specificity. Malted AI was started based on our founders winning the Amazon Alexa Challenge, where they leveraged Small Language Models (SLMs) and distillation techniques to solve domain-specific tasks in a secure production environment. These learnings are the cornerstone of Malted AI's unique approach to helping enterprises extract value by focusing on specificity instead of generalism. Check out our first blog post on how the Alexa Challenge inspired what Malted AI has become today! 👇 https://lnkd.in/eF8SASpV

We're thrilled that Malted AI has been featured as one of the top 5 innovative OpenAI rivals emerging from Europe in 2024. Our approach of combining Small Language Models (SLMs) and distillation allows enterprises to solve complex problems where general AI fails. We will continue to push the boundaries of what is possible with smaller, more focused AI models. 🌾💥

🤖 While tech giants like OpenAI dominate headlines, a new wave of European startups transforms the field with innovative approaches and niche solutions, with the UK leading the charge. 🤖 A new generation of startups is emerging, each with unique approaches to AI challenges. Meet 5 innovative OpenAI rivals emerging from Europe in 2024: Malted AI - Iain Mackie Sana - Joel Hellermark AI Sisters - Emma Pariente Chloé Pariente Nordics.io - Lukas Rezanina Robert Decman GoodAI - Marek Rosa #AI #startup #tech #funding #news #EUtech #EUstartup #innovation

We had an incredible time at Turing Fest and were thrilled to have our CEO and Co-Founder Iain Mackie give a keynote speech. Iain shared how winning the Amazon Alexa Challenge helped shape our technology thesis and our progress in distilling Small Language Models (SLMs) for enterprise. Some highlights from Iain's talk: * High-performing machine learning models need 3x things: ‘data, data, data’. * Distillation is a game changer in creating scalable, high-quality data. * Small language models (SLMs) are a fantastic choice for solving specific, high-value problems in production environments (optimising economics, latency, and performance). * Networks of SLMs working together can outperform LLMs, while being more cost-effective and able to deployed on a a small GPU. * Scotland is a great place to start an AI startup, given the amazing software and AI talent. Don't hesitate to reach out if you want to learn more about small language models. Big thanks to Brian Corcoran and Turing Fest team for the great show. Federico Rossetto | Carlos Gemmell I Laura Bernal Vergara  

We had an incredible time at Turing Fest and were thrilled to have our CEO and Co-Founder Iain Mackie give a keynote speech. Iain shared how winning the Amazon Alexa Challenge helped shape our technology thesis and our progress in distilling Small Language Models (SLMs) for enterprise. Some highlights from Iain's talk: * High-performing machine learning models need 3x things: ‘data, data, data’. * Distillation is a game changer in creating scalable, high-quality data. * Small language models (SLMs) are a fantastic choice for solving specific, high-value problems in production environments (optimising economics, latency, and performance). * Networks of SLMs working together can outperform LLMs, while being more cost-effective and able to deployed on a a small GPU. * Scotland is a great place to start an AI startup, given the amazing software and AI talent. Don't hesitate to reach out if you want to learn more about small language models. Big thanks to Brian Corcoran and Turing Fest team for the great show. Federico Rossetto | Carlos Gemmell I Laura Bernal Vergara