Recogni’s Post

New open-source model, FalconMamba 7B, from Technology Innovation Institute with the Mamba architecture just dropped! Just a reminder that you still have lots of exciting research and innovation happening at the model architecture level. State Space Language Models can handle longer sequences of text more efficiently, require less memory to operate, and maintain consistent performance regardless of input size. Read the article on Maginative - https://lnkd.in/eVvXsWKQ

As AI solutions evolve, the transition from a single-agent system with tool calls to a multi-agent architecture feels like a natural next step for tackling more complex challenges. For those who’ve made this leap, what was the tipping point that led you to embrace multi-agent systems? I’d love to hear about the key factors and experiences that guided your decision!

Sometime next week I will be posting an article here and on my substack about redefining the next step of AI by ditching the currently used term (and associated metaphor) - AGI. This is a good time to challenge assumptions (not only about the limitations of GenAI that have resulted due to its unsustainable footprint) but also how to position ourselves conceptually to move out of the current roadblock to a more deliberate architecture.

Everest Group is leading the charge in the ever evolving #AI landscape. Explore this detailed visual breakdown of the #AgenticAI architecture, designed to revolutionize industries by empowering intelligent systems to think, act, and evolve independently. Have a look: https://okt.to/TF2jkx Curious to learn more about how this cutting-edge technology can transform your business? 👇 Go for a deeper dive into this research: https://okt.to/vptlIn Get in touch: @Anil Vijayan @Vaibhav Bansal @Vershita Srivastava @Divya Chandak @Shreepriya Sinha

Our President, Peter Carson, is at the DIALOG AI & Partner Conference today. He was invited as an industry expert to lead a discussion on advancements in AI and how they relate to the Architecture, Engineering, and Construction industries. The future of AI is happening now, and we're thrilled to be part of the conversation! #AI #AIInnovation #TechPartnerships #FutureOfWork

Which architecture makes more intelligent AI agents: Graph RAG or Vector RAG? Our team at Tars trained two AI Agents on both architectures using information from Harry Potter and The Prisoner of Azkaban. Witness the spell-binding differences that emerge in this video! We are developing a webinar that will explore this topic in depth. If you are interested, you can sign up here. https://lnkd.in/gk3Jaw3G

Everest Group is leading the charge in the ever evolving #AI landscape. Explore this detailed visual breakdown of the #AgenticAI architecture, designed to revolutionize industries by empowering intelligent systems to think, act, and evolve independently. Have a look: https://okt.to/ot6pRq Curious to learn more about how this cutting-edge technology can transform your business? 👇 Go for a deeper dive into this research: https://okt.to/eAZ5D2 Get in touch: @Anil Vijayan @Vaibhav Bansal @Vershita Srivastava @Divya Chandak @Shreepriya Sinha

Recently, I’ve observed an intriguing trend in my discussions with customers—certain key concepts are consistently surfacing that are crucial for our future success. Terms such as Agentic Architecture, Multi-Agent Framework, Knowledge Fabric, and Knowledge Graph have become increasingly prominent in our conversations. These concepts are not merely industry jargon; they represent the innovative landscape we are beginning to explore. Grasping these ideas is vital for advancing our strategies and ensuring we are well-prepared for the challenges and opportunities that lie ahead. Understanding their implications will help us navigate our path toward innovation and efficiency in an ever-evolving market. #GenerativeAI #GenAI #AgenticArchitecture #MultiAgentFramework #KnowledgeGraph #AIStrategy

If you watch this video, you'd conclude that Human Intelligence (𝐇𝐈) won over 𝐀𝐈. If so, you will be... 𝐜𝐨𝐦𝐩𝐥𝐞𝐭𝐞𝐥𝐲 𝐰𝐫𝐨𝐧𝐠. This is about Tom & Jerry, the cat & mouse. The reason it's very wrong is that it's not about HI vs AI but about the game. 𝐇𝐞𝐫𝐞'𝐬 𝐚𝐧 𝐚𝐥𝐭𝐞𝐫𝐧𝐚𝐭𝐢𝐯𝐞 𝐩𝐨𝐢𝐧𝐭 𝐨𝐟 𝐯𝐢𝐞𝐰. This is classic offense - defense interplay. ➜ The defense develops a strategy, and it works for some time.... ➜ After a number of failed attacks, attackers learn what does not work ➜ Attackers try new approaches until somebody stumbles on one that works. ➜ And then the defense adapts, and similar attacks fail. ➜ 𝐀𝐧𝐝 𝐭𝐡𝐞 𝐜𝐲𝐜𝐥𝐞 𝐫𝐞𝐩𝐞𝐚𝐭𝐬 𝐢𝐧𝐝𝐞𝐟𝐢𝐧𝐢𝐭𝐞𝐥𝐲. 𝐘𝐨𝐮 𝐰𝐢𝐥𝐥 𝐬𝐞𝐞 𝐭𝐡𝐢𝐬 𝐜𝐲𝐜𝐥𝐞 𝐢𝐧 𝐦𝐚𝐧𝐲 𝐟𝐢𝐞𝐥𝐝𝐬: military history, chess, sports, physical and cyber security, and corporate strategies. It has nothing to do with human intelligence (HI) vs AI. HI vs HI has the same pattern. 𝐈𝐭'𝐬 𝐚𝐛𝐨𝐮𝐭 𝐭𝐡𝐞 𝐧𝐚𝐭𝐮𝐫𝐞 𝐨𝐟 𝐭𝐡𝐞 𝐠𝐚𝐦𝐞, 𝐧𝐨𝐭 𝐫𝐞𝐚𝐥𝐥𝐲 𝐭𝐡𝐞 𝐧𝐚𝐭𝐮𝐫𝐞 𝐨𝐟 𝐢𝐭𝐬 𝐩𝐥𝐚𝐲𝐞𝐫𝐬. 𝐌𝐲 𝐫𝐞𝐚𝐥 𝐩𝐨𝐢𝐧𝐭: This is a great post by Grant Ecker. Not because it shows that Humans are smarter than machines. It's a brilliant post because we can use it to make sure we ask the right questions AND avoid the right answers to the wrong question. P.S. The robot goalie may not necessarily use AI. But it does not matter. The analysis above holds in any case. #AI4Leaders #AI #HI #Strategy #Security

Curious about the future of open computing? Catch Krste Asanovic’s RISC-V State of the Union from the recent RISC-V Summit! Krste dives into adoption trends across markets, ecosystem programs fueling growth, and new extensions advancing AI applications. Watch the full keynote here: https://hubs.la/Q02XY9Tc0