Propheus

We’re hiring an AI Engineering Lead! At Propheus, our E.D.G.E. team works at the intersection of AI and business expertise - delivers real ROI for enterprises on the ground. If you believe the future belongs to engineers who go deep into real-world problems and solve them with tech, we want you. Apply via the link in comments or email sachin@propheus.com

We're #hiring a new AI Engineering Lead in Bengaluru, Karnataka. Apply today or share this post with your network.

When AI Meets Kindness: Propheus as the Solution for Stray Dogs. The recent Supreme Court ruling on stray dogs has ignited intense debate by focusing primarily on sheltering the animals — a move many argue is neither practical nor a complete solution. The factors that allow stray populations to grow - from poor sanitation to high-density food vending zones - are deeply embedded in the way our cities function. What is clear is that we need a smarter approach that protects people without abandoning compassion for animals. This is where AI and technology could make a real difference, offering targeted, data-driven interventions instead of blanket measures. Here is how, at Evam Labs, we've solved the problem using our flagship product, Propheus. Check out this article in the comments to know how! Shobhit Shukla Sumanth N Sreejan Choudhary Sachin Jose Varghese

Atlas is Reborn. No longer burdened by the weight of the world, he walks it — free, agile, and hyperaware. Powered by his AI Agents, he decodes the pulse of the planet: places, mobility, traffic, weather, sentiment, news. He doesn’t just carry the Earth. He understands it. This is Physical AI. Get a glimpse of Atlas Reborn at the link in the comments. Shobhit Shukla Sreejan Choudhary Rahul Agarwal Sachin Jose Varghese AswinPratap Narayanasamy Michelle Zhou Saurav Bharadwaj Elishia Vaz Sumanth N Abhijit Hubli Diksha Rudrawal Vibhav Sharma Prathamesh Mayekar Vishaal Rao Mithul Chander

Behind the scenes of the Digital Atlas — our foundational layer for understanding the physical world through multimodal intelligence. It powers real-world reasoning for Agentic AI systems, enabling use cases like site selection, demand forecasting, market growth potential and much more. 👇 Watch the video to see how we’ve brought this vision to life.

In our Alchemy Benchmarking process, we go beyond surface-level comparisons — validating not just what the models say, but how they think. Our two-tiered validation system checks:   👉 Agent Pathways — Is the model producing a valid, executable query that aligns with the business intent?   👉 Answer Accuracy — Does the final output match ground truth? 🔗 Dive deeper into our validation framework in the blog link attached in the comments.

Unlock high potential growth zones — before your competitors do. 🥁 Drumroll, please... Introducing The Growth Potential Agent — Powered by dynamic real-world signals, it helps you pinpoint high-opportunity zones, uncover what’s driving demand, and plan smarter, faster expansions. 🔗 Try the demo — link in comments.

AI is evolving—from raw queries to contextual intelligence. Alchemy is built for the enterprise. A multi-agent ecosystem designed for real-world business questions. ⏩ Stay tuned — the Alchemy sandbox is launching soon.

Alchemy transforms raw tables into higher-order business concepts. For example — a good AI system should know what “store performance” means in context — not just pull numbers from a database. It should factor in location, foot traffic, store size, and customer behavior — just like your teams would. Instead of simple table joins, concepts create higher-order representations that combine key attributes, behavioral indicators, and contextual logic. This semantic approach enables AI systems to understand business context — factoring in location, foot traffic, and customer behavior — rather than just executing SQL queries. 📊 The result: more accurate, relevant insights for critical use cases like customer segmentation, trade area analysis and campaign targeting. 🔗 Check how concepts improve accuracy—Alchemy Benchmarking blog linked in comments.

📊 From our Alchemy benchmarking results: Open-source models are rapidly closing the gap — and in some cases, even outperforming — closed-source counterparts on factual and pattern-based retail enterprise queries. As shown below, Qwen 3–235B outperforms Claude Sonnet-4 on these tasks. What does this mean for enterprises? It unlocks the potential to cut costs, increase flexibility, and maintain top-tier performance — without relying solely on proprietary systems. 🔗 Full results in the blog (link in comments)