We’re kicking off 2026 with a fresh set of agent-building updates, improved experiment comparison, and new reads on observability and evaluation. Read on to see everything we’ve shipped this January.

Product Updates

LangSmith

🚀 LangSmith Agent Builder is now GA

Build agents with natural language. Describe what you want, and Agent Builder figures out the approach, including a detailed prompt, tool selection, subagents, and skills. Try Agent Builder.

📊 Spot regressions and improvements at a glance with side-by-side LLM experiment comparisons. Filter by inputs, outputs, status, or metadata to zero in on what matters. Learn more.

Open Source

LangChain JS v1.2.13 improves agent robustness with dynamic tools, recovery from hallucinated tool calls, and better streaming error signals. Learn more.

Stream live progress from subagents and surface who’s doing what as messages are generated in deepagents. Learn more.

Interrupt 2026 - Tickets on sale Feb 12

Join us May 13-14 for Interrupt 2026, the AI Agent Conference. Tickets drop February 12 and are limited— early registration recommended. Sign up to hear first when sales open.

Speak the Lang

Agent observability powers agent evaluation

Traditional software treats tracing and testing as separate. With agents, they’re inseparable. Teams shipping reliable agents use a single workflow where production traces directly power their evaluations.

Why this matters:

🔹 Agent behavior only emerges at runtime—traces show what actually happened

🔹 Evaluating agents means checking trajectories, outputs, and state, not just final answers

🔹 Production traces become living test cases that shape what you evaluate next

Read more here on how traces become the foundation of an agent eval strategy.

Agent Builder’s memory is a filesystem

Agents doing repeated tasks need to remember, so we gave Agent Builder memory using standard Markdown and JSON files. Here’s how we built it and what we learned. Read more.

New Agent Builder Academy Course

Learn how to build agents with LangSmith Agent Builder. We’ll cover set up, building and improving your agent, and then dive into Triggers, Subagents, MCP, and Skills. Learn more.

Upcoming Events

Upcoming Events

🇨🇳 (Feb 1) Shanghai // Community Meetup: Agent Builder Meetup.

Get an in-depth intro into LangSmith Agent Builder and Deep Agents, organized by Ambassador Haili Zhang. RSVP here.

🌐 (Feb 5) Virtual // Webinar: Agent Observability Powers Agent Evaluation

Walk through the core primitives of agent observability and learn how teams use them together to improve agent behavior. RSVP here.

🏙️ (Feb 5) NYC // No-Code Agent Building (Women & Underrepresented Genders)

Calling all citizen developers, technical professionals, and non-technical folks who are women and gender-divers - come build an agent with us! RSVP here.

🗽(Feb 17) New York // AI Agents Meetup: Agent Observability Powers Agent Evaluation

Join our Nick Huang, Applied AI Engineer at LangChain, as he shares best practices on agent observability & evals with LangSmith. RSVP here.

🌉 (Feb 18) San Francisco // AI Agents Meetup: Agent Observability Powers Agent Evaluation

Join our CEO, Harrison, as he shares best practices on agent observability & evals with LangSmith. RSVP here.

🇮🇳 (Feb 21) Bengaluru // Community Meetup:

This meetup will focus on building agents with eval-driven auto-optimization and building a universal memory layer for AI. Organized by Ambassador Ravi Kiran Vemula. RSVP here.

🇳🇱 (Feb 25) Netherlands // Community Meetup: Utrecht

Hear from the LangChain team on how to build reliable agents with LangSmith, alongside Karsten (Field CTO at Incentro). RSVP here.

🇸🇪 (Feb 26) Paris // Community Meetup: Agents & Apéro

Ambassador Juan Felipe Arias and and Ambassador/Expert Guillaume Fortaine. keep things informal, friendly, and very Parisian — with an apéro vibe and plenty of time to network 🍷🥖. RSVP here.

🇺🇸(Feb 26) Denver // Community Meetup: Detecting Errors Early with LangSmith

Joined Focused's Lead Agent Engineer for a practical discussion on building and operating production-ready AI applications with LangSmith. RSVP here.

🇸🇪 (March 22) Stockholm // Community: Lovable x LangChain Proptech Hackathon

Join LangChain and Lovable for a one-day hackathon in Stockholm. Build AI solutions for Real Estate & Construction using the next gen of AI development tools. Organized by Ambassador Gustaf Gyllensporre. RSVP here.

🇳🇱 (March 26) Amsterdam // Community: The Rise of Autonomous Systems

Get practical POVs on architectures, runtime, and the evolving role of platforms in enabling reliable AI at scale — with speakers from LangChain, AWS, Qodo, and SurrealDB. Organized by Ambassador Sri Rang. RSVP here.

🤝 Customer stories

• Coinbase standardized a code-first, observable agent stack to safely automate regulated workflows — reducing agent development time from quarters to days. Read the full story.
• Remote built a Code Execution Agent that separates reasoning (LLMs) from execution (Python), turning complex, multi-format employee and payroll data into validated JSON in hours. Read the full story.

How can you follow along with the Lang Latest? Check out the LangChain blog, Changelog, and YouTube channel for more product and content updates.