LangChain

What do Ramp, Perplexity, Superhuman, and Replit have in common? They've built breakout agentic apps that are live in production and pushing the boundaries of what AI agents can do 🤩🚀✨ In our new "Breakout Agents" series, we've teamed up with Ramp, Perplexity, Superhuman, and Replit to take you behind-the-scenes of how they built their AI agents. 🎬 Explore how their engineering teams: • Designed UXs for human-agent interactions • Built unique cognitive architectures • Leveraged prompt engineering • Evaluated app performance to stay ahead ➡️ Read their stories and get inspired: https://lnkd.in/g246xsH8

✒️Kiroku is a multi-agent system that helps you organize and write documents Really complex agent (see the diagram below!) that HEAVILY involves a "human in the loop" flow Great resource for anyone looking to create a writing agent https://lnkd.in/gz9ciyye

📊Implementing GraphReader with Neo4j and LangGraph Elevating RAG accuracy and performance by structuring long documents into explorable graphs and implementing graph-based agent systems From our friends at @neo4j https://lnkd.in/gzEJWdpE

🧱LangChain and Elasticsearch accelerate time to build AI retrieval agents A few weeks ago we launched LangGraph Templates, including one that sets up an agent to do retrieval Easiest way to get started! Code: https://lnkd.in/gtBjZcVB Press release: https://lnkd.in/gW-vsD3f

✨SCIPE✨ - Systematic Chain Improvement and Problem Evaluation TLDR: It helps you find underperforming nodes in LLM chains New research out of Berkeley (Shreya Shankar and Ankush Garg) that integrates seamlessly with LangGraph "In this post, we introduce SCIPE, a lightweight, yet powerful tool that conducts error analysis on LLM chains. This tool can benefit anyone creating applications that rely on LLMs for making decisions and carrying out tasks. SCIPE works by analyzing inputs and outputs for each node in the LLM chain and identifying the most important node to fix–the node that, if accuracy is improved, will most improve the final or downstream output accuracy." Blog: https://lnkd.in/g_e6_2Z4 Collab Notebook: https://lnkd.in/gaTPnjUv

New to LangSmith? Check out our 7-part video series that walks through the basics of LangSmith for LLM testing & observability. Learn how to: 🔍 Set up tracing for visibility into your LLM app 🛝 Build & refine prompts in the Playground 🛠️ Curate datasets & run robust evaluations ✅ Use annotation queues for human feedback 📊 Run online evaluations to score performance in real-time 📈 Monitor LLM app health with customizable dashboards Watch now: https://lnkd.in/dcy4PBPz

🧑⚖️How Chaos Labs built a multi-agent system for resolution in prediction markets Prediction markets rely on an "oracle" to determine outcomes and resolve bets. Read how Chaos Labs built a multi-agent research system to do this in an automated way https://lnkd.in/gYGuAVVx

Join LangChain at ElasticON NYC on November 13th for talks on how AI-powered search is driving real impact. 🔍 🎙 Catch Vadym Barda in a panel discussion on how search AI is transforming app development and how we're using Elastic to redefine customer experiences. 🎓 Join Nick Huang as he dives into building production-ready RAG apps by combining vector DB capabilities with traditional search approaches and secure data retrieval. Register now: https://lnkd.in/grzDTzev

🤖Using remote graphs in a multi-agent setup Connect to any deployed graph with our new RemoteGraph SDK. Makes it possible to use a remotely running graph as part of a multi-agent architecture Docs: https://lnkd.in/gwWqdBEk Video: https://lnkd.in/g4KQv6VG

🕑From 12 weeks to 10 minutes: How Novo Nordisk Accelerates Time To Value with GenAI Great case study on how NovoScribe, a RAG solution, was built with LangChain and MongoDB With LangChain "the team can switch between models quickly and easily" https://lnkd.in/gTh-QTpC