Holistic AI

Listen to hear why we HACK… Raj Bharat Patel, CQF Emre Kazim Eleanor Johnston Seonglae Cho #TheGreatAgentHack #HolisticAI #UCL #AWS #NVIDIA #MAPFRE #PGIM #RAIUK #AgenticAI #AIHackathon #AIGovernance

Join us behind the scenes at The Great Agent Hack 2025 for a glimpse into the attendee experience. #TheGreatAgentHack #HolisticAI #UCL #AWS #NVIDIA #MAPFRE #PGIM #RAIUK #AgenticAI #AIHackathon #AIGovernance

Why did we host the Great Agent Hack 2025? Because AI is entering a new era – one where agents don’t simply generate; they make decisions, take actions, and shape real-world outcomes across the enterprises that rely on them. And that requires agents that are strong, transparent, and trustworthy. That’s what makes this hackathon different. It’s not just a weekend of coding. It’s a community of researchers, engineers, developers, and industry leaders who understand that building agents for enterprise deployment demands rigor, humility, and accountability. At Holistic AI, we turn deep experimentation into better, safer tools for enterprises building high-impact systems. The Great Agent Hack is one way we put that ethos into practice by bringing industry and academia together to test boldly, surface risks early, and define what “good” should look like for agentic AI. Because safe, high-performing AI doesn’t happen by accident. It’s built by people who care – and who understand the responsibility of building systems that will be trusted in business-critical environments. Check out the video for a glimpse of the energy that defined the weekend. #TheGreatAgentHack #HolisticAI #AgenticAI #AIHackathon #UCL #Valyu #AWS #NVIDIA #PGIM #MAPFRE #RAiUK

Today we’re excited to spotlight the award winners from the Great Agent Hack 2025! We handed out 26 awards in total, including the Grand Champion, first-place winners from each track (focused on agent performance, visibility, and safety), the Public Choice Award, and the Brave Soldier Award — presented to an individual who pushed through and completed the challenge on her own after the rest of her team was unable to continue after the first day. The projects demonstrated impressive technical depth, thoughtful experimentation, and genuine creativity. A selection of winners is highlighted below, and the full list is now live on the Great Agent Hack 2025 site → https://lnkd.in/eUqRhSjS Congratulations to all of our award recipients! Thank you for your bold ideas and commitment to building higher-performing, more transparent, and more reliable agents. #TheGreatAgentHack #HolisticAI #AgenticAI #AIHackathon #AIGovernance #UCL #Valyu #AWS #NVIDIA #PGIM #MAPFRE

After submitting their projects, teams stepped up to present — and they delivered. The sessions showcased real creativity, sharp problem-solving, and thoughtful approaches to the biggest challenges in agentic AI. Judges spent the afternoon moving from screen to screen, digging into each team’s framing, design choices, and implementation details across all 51 submissions. The review was intense and inspiring, and it even led to a brand-new mystery award created on the spot for a team that simply had to be recognized. Tomorrow we’ll reveal who earned that surprise honor, along with the full list of award winners — including the Grand Champion of the Great Agent Hack 2025! #TheGreatAgentHack #HolisticAI #UCL #Valyu #AWS #NVIDIA #PGIM #RAIUK #MAPFRE #AgenticAI #AIHackathon #AIGovernance

Huge thanks to our partners and sponsors for powering the Great Agent Hack 2025! This hackathon was a true collaboration, and it simply wouldn’t have happened at this scale without their support. We’re grateful to Amazon Web Services (AWS), NVIDIA, Valyu, PGIM’s RealAssetX, Responsible Ai UK, MAPFRE and  UCL Centre for Digital Innovation (CDI) and UCL Centre for Sustainability & RealTech Innovation (CSRI) — along with Swift Food UK, and all of the judges, mentors, and organizers who contributed their time, expertise, and resources throughout the weekend. Together, we created a space where over 200 participants could experiment with agentic AI in a real-world setting — and we’re excited to build on this momentum for next year’s Great Agent Hack. #TheGreatAgentHack #HolisticAI #UCL #Valyu #AWS #NVIDIA #PGIM #RAIUK #MAPFRE #AgenticAI #AIHackathon #AIGovernance

The Great Agent Hack 2025 is officially a wrap — and what a weekend it was. Over 200 builders came together at UCL East to explore real-world challenges in agentic AI across three tracks: performance, visibility, and safety. From early-morning builds and mentorship conversations to packed poster sessions in the afternoon, the energy in the room never dipped. Teams experimented, stress-tested ideas, and pushed the boundaries of what agentic systems can do. This is just a snapshot of the weekend. In the coming days, we’ll be highlighting selected projects, insights, and standout moments from across the hack — including our award winners and follow-on collaborations. Huge thank you to everyone who took part and helped bring the Great Agent Hack 2025 to life! #TheGreatAgentHack #HolisticAI #UCL #Valyu #AgenticAI #AIHackathon

🚨 Is There a Shift East in the Global AI Landscape Coming? A Chinese Open-Weight Model Just Matched (or Beat) Claude & GPT in Safety Tests A new Holistic AI red-team analysis just dropped and it could be one of the most consequential signals yet that the AI race is entering a new phase. 🔍 Holistic AI’s evaluation of leading Chinese open-source/open-weight models reveals something few expected: 👉 MiniMax’s M2 (Thinking) model performed on par with, and in some categories better than, Claude 4.5 and GPT-4.5 on safety and jailbreak resistance. 🧨 Better safety from a Chinese open-weight model. 💰 At ~8% of the cost. Is this a blip or a turning point? Zeus Kerravala explains in TechRepublic. #OpenSourceAI #ResponsibleAI #TrustedAI #ChineseLLMs #RedTeaming #Jailbreaking https://lnkd.in/g-xPBRbZ

🚀 It was incredible to be in Barcelona last week for Mozilla Festival, where the theme was rethinking power and imagining alternatives in tech   Emre Kazim joined Philip Dawson, Anna Spitznagel and Camilla de Coverly Veale to discuss a key question: how do we unlearn AI to foster responsible development? This session was a reminder that building responsible AI requires a collaborative, sociotechnical approach, where AI governance is integrated throughout the entire AI lifecycle 🤝   Huge thanks to Saayeli Mukherji Bruni, Mohamed Nanabhay, and the whole Mozilla Ventures team for having us and bringing together this community to make these conversations possible!   Already looking forward to next year! 🙌   #Responsibleinnovation #UnlearningAI #Mozfest #AIGovernance Ayesha Gulley Airlie Hilliard

Day 2 at the Great Agent Hack 2025 is in full swing and 51 submissions are officially in! Teams are now presenting posters across three tracks — performance, visibility, and safety — with an incredible range of ideas, experiments, and build approaches. In just two hours, we’ll be awarding $65,000 in prizes across more than a dozen categories. Who’s going to take home the wins? We can’t wait to find out.