Tavily

We're hosting a game night for tech founders in NYC! Tavily, The Collective, MongoDB, Redis & Eden are bringing together a group of developers and tech founders for a night of poker, chess, backgammon, and more! RSVP to join us: https://luma.com/u85izm5b 🎰

I had a blast joining Srinivas Pendyala and Udayasimha Theepireddy on stage at AWS re:Invent to talk about context engineering and building better agents. A few takeaways I’m still thinking about: • Your proprietary data is your moat. If you can’t turn it into usable context, you lose the game. • Context window ≠ RAM. Redundant data costs more and can increase hallucinations. The goal is precision context, not maximal context. • No off-the-shelf context stack will fit you. Build your own internal retrieval, and leverage third-party MCPs for external knowledge. Huge thanks to Amazon Web Services (AWS) for having us, and to everyone who came by to chat about Tavily after the session. Excited for what this space looks like in 2026.

Our first AWS re:Invent is officially wrapped, and we’re heading home energized — by the progress in retrieval and agent reliability, and by the meaningful connections we made across the builder and partner community. 🔵Our Founder + CEO, Rotem Weiss, took the stage with Amazon Web Services (AWS) and Elastic to share practical insights on context engineering and what it really takes to build reliable AI agents. 🔵We were thrilled to see Tavily powering demos across the expo floor from Redis, LangChain, JetBrains, and Weights & Biases. Huge thanks to these teams for showcasing just how powerful strong retrieval can be. 🔵And in conversations with builders, founders, and enterprise teams, one theme was impossible to miss: AI needs grounding and retrieval is becoming critical. We’re already looking forward to coming back next year with even more to share. Onward⚡️

👀 In case you missed our announcement last week, thousands of developers and teams have already shown interest in our new /research endpoint — and we just published a full deep dive on Hugging Face that breaks down how it works and why it’s different. If you want to understand the technical and philosophical lessons behind building a state-of-the-art research, this is for you. 📘 Read our new blog on Hugging Face: https://bit.ly/44NOy1w 🔍 Learn more and sign-up here: https://bit.ly/44ul4WH The excitement so far has been incredible. Don’t miss what everyone’s exploring.

At Tavily, we believe great agents start with great information. That’s why we’re excited about how seamlessly LangChain brings Tavily into an end-to-end research workflow. From querying the web to delivering actionable insights. Awesome to see this come together.👇 Dive into the demo below.

✅ Day one in the books for the Tavily team at AWS re:Invent! We spent yesterday running demos across JetBrains, LangChain, and Weights & Biases, showing how sharper retrieval turns “agents” into something teams can actually rely on. Lots of great conversations, lots of “this is exactly what we’ve been missing” moments. Today’s the big one: Rotem Weiss joins Amazon Web Services (AWS) and Elastic at 5:30 PM to break down practical context-engineering techniques for building agents with genuinely relevant, trustworthy retrieval. If you care about making agents production-ready, don’t miss it. We’ll also be around the Expo floor throughout the day. We'll be with Redis at 10:45 AM and Weights & Biases at 3 PM. Come find us and say hello!

JetBrains set out to enhance their AI Assistant with accurate, up-to-date knowledge from the web and developer docs without slowing teams down with heavy engineering work. They turned to Tavily, integrating our API directly into their IDEs to deliver: ✅ Fresh, reliable answers ✅ Scoped documentation search (PyTorch, NumPy, and more) ✅ Clean formatting and safe search out of the box As one JetBrains engineer said: “With Tavily, it just worked.” Grab some coffee and ☕️ read the full case study: https://bit.ly/48hWtFp And if you're at AWS re:Invent this week, you can see it live. Join Tavily at the JetBrains Booth (#1385) on Dec 2, 10am–12pm, where Dean Sacoransky will demo how the JetBrains coding agent uses Tavily to level up developer workflows with real-time intelligence.

🎟️Tavily is hosting its first-ever digital event and you’re invited. Not a blog post. Not a tweet thread. A real, live session where we finally show you what we’ve been building in action. We’re calling it Tavily 101 — a chance for our developer community to come together around one big question: “How do you actually plug LLMs into the real-time web?” On December 10th at 12pm ET, the Tavily team will walk through exactly that. No abstractions. No mystery. Just a hands-on look at how Tavily works and how teams are using real-time search to level up their AI agents, copilots, and applications. What to expect: 🔵How the Tavily API works 🔵Real examples pulled directly from production use cases 🔵How developers integrate Tavily in minutes 🔵An open Q&A. Jump in, ask anything, drop comments, and build with us It’s our first time doing something like this, and we’d love to create it with you. Join us for our first digital event: https://bit.ly/4oduXPh

🚀 Tavily’s new /research endpoint is live and it just hit #1 on DeepResearch Bench. This is a huge milestone for the team and a signal that deep, source-grounded, real-time research can finally be both fast and reliable. We’re opening Early Access to our new out-of-the-box research agent, designed to deliver: 👉Deep, structured, verifiable insights 👉State-of-the-art performance and efficiency 👉No agent-building required Check it out here → https://lnkd.in/grvcb7QS