Tecton’s Post

AI’s productivity gains aren’t coming at the expense of devs 💪 Here’s how they’re using AI tools to free up capacity to drive impact at speed 👇 If you want more data, and less opinions on AI’s impact on the ZA tech ecosystem, get our latest AI Skills and Impact Report created with the tech community 👉 https://bit.ly/4bbmYf5

Earlier this week I spoke about one of the hottest topics in AI, Knowledge graphs! Massive thanks to the organizers of AI Wednesdays, Smart Nation Singapore, for the amazing opportunity. Most of the tooling around knowledge graphs revolves around neo4j + langchain. That’s a lot of code and abstraction between you and your graph. My presentation was about building a knowledge graph extraction pipeline from scratch, in Rust. This sort of mental exercise to go one level down always helps to clarify which steps are critical, and what can be trimmed away. Technical points in the comments!

We’re excited to share that #DatabricksVentures has invested in the Series B of Galileo 🔭, a startup focused on Evaluation Intelligence for AI teams everywhere! One of the biggest challenges to getting GenAI into production is the “measurement problem” - the difficulty of evaluating performance in real-world settings. Galileo empowers teams with scalable evaluations that span entire AI workflows, work out-of-the-box, and continuously improve. All Databricks models are natively available to Galileo users. https://dbricks.co/4hiNHKT

🎉 Exciting times at Corsound AI! We’ve just wrapped up our second session in the City of London Corporation and Microsoft innovation challenge and the journey keeps getting more insightful. As we dive deeper, it’s becoming clearer how vital deeptech is in the fight against deepfakes. Here are some key takeaways from today's session: 💡 1. Deepfakes: A Cross-Domain Challenge We witnessed a range of innovative solutions across different fields—from document verification to voice and video analysis. Each one tackling the deepfake threat in its own unique way, highlighting the complexity of the problem. 📈 2. The Rapid Rise of Deepfake Threats Everyone agreed: deepfake crimes are on the rise. With GenAI tools improving rapidly, the need for solutions like ours has never been more urgent. 🛠 3. Keeping Pace with Generative Tech A hot topic was how to stay ahead of ever-evolving generative AI. Some companies are partnering with these AI tool providers to access and train models directly on their platforms—a smart approach to outpacing the threat. 🔜 What’s Next? * Continue learning and sharing insights with our fellow innovators. * Deepen our understanding of evolving deepfake challenges in the finance sector. * Strengthen partnerships with financial companies, offering our voice/voice-face match deeptech to solve real-world problems. We’re excited to see where this journey takes us next! 🚀 #innovation #AI #deepfakes #genAI #fintech

One of the most critical aspects of every product decision we make is validation—and it's especially vital for young deeptech startups like ours that are tackling emerging threats, like GenAI deepfakes. 🏦 The London Innovation challenge , gives us an incredible opportunity to walk alongside some of the biggest names in the financial world, including Mastercard and HSBC, validating our use cases, refining our business flow, and receiving invaluable market feedback. 😎 What’s even more exciting? We’re not doing this alone. We're part of a diverse forum of companies, from early-stage startups to more mature tech solutions, all united by a shared mission: to fight deepfakes. Each of us is bringing a unique perspective to the table, and the collective expertise—paired with technological support from Microsoft—is priceless. But that’s not all. We believe that the insights we gain from this journey shouldn’t stay behind closed doors. We want to share our progress with you! 💡 Each week, we’ll be posting updates about our learnings, feedback, and breakthroughs, so that our community can grow and improve together. We’d also love to hear your thoughts, dilemmas, and ideas on tackling the challenge of deepfakes. Let’s work together towards a future with stronger security and trust in the digital world. 🔒💬 Stay tuned, and let’s keep the conversation going! #innovation #deepfakes #AI #community #LondonInnovationChallenge

Shining a light on an excellent recent client commentary win! In an article with Cointelegraph, ChainGPT's founder and CEO Ilan Rakhmanov shares his thoughts on the prominence of big tech players in the AI boom and the potential impact their new tools could have on the industry. Congrats to Ilan on the great insights! Find the full article here >> https://lnkd.in/depZd2v6

AI’s shaking up dev skills and tech assessments 🤖 Here are 3 pro-AI strategies you can use to get the best understanding of candidates’ skills 👇 If you want more data, and less opinions on AI’s impact on the ZA tech ecosystem, get our latest AI Skills and Impact Report created with the tech community 👉  https://bit.ly/4bbmYf5

Excited to share a brand new episode of our GenAI's Lamp YouTube series! 🧞 In this episode, we discover the power of Hugging Face's Models, Endpoints & Spaces, and learn to maximize their potential for your AI applications! 🔍 What's Inside: - Overview of interacting with state-of-the-art models from Hugging Face - Demonstration of making API calls directly to a model by using Hugging Face libraries in Node.js - Exploration of embeddings and feature extraction with different models - Use of LangChain framework for Hugging Face inference and embeddings - Implementation of the Retrieval Augmented Generation (RAG) pattern - Hosting Flowise on Hugging Face and creating a ChatFlow with conversational Q&A chain Watch now and unlock a world of possibilities! #HuggingFace #Nodejs #LangChain #Flowise #TechAtWorldline https://lnkd.in/euQWw5Yb

Excited to share a recent demo on conversational AI, showcasing the potential for voice agent development in 2025. The technology is advancing rapidly, and we’re seeing great promise in building scalable solutions. Key highlights of the PoC: - Knowledge Base: Powered by Neo4j graph database - Toolkit: LangGraph multi-agent system wrapped as a API endpoint - Model: OpenAI with real-time voice capabilities Looking forward to exploring more use cases and pushing the boundaries of AI-driven voice interactions!

Knowledge Workbot Accelerator from Workato 🌠 This Accelerator leverages generative AI with supported LLM providers and Vector databases to generate answers from diverse data sources to provide seamless experience to the users without needing to go to multiple data source systems and find correct answers. This accelerator follows a modular approach to enable easy replacement of any of the systems in the accelerator with speed and ease.