Channel
Digests from LangChain
Building self-improving agents requires understanding three parts: Model, Harness, and Context. The foundation for improvement is robust observability via Traces and Evals.
Read article
Managed Deep Agents streamlines the journey from local agent prototype to production. It couples the agent harness with LangSmith infrastructure, simplifying complex tasks like durable execution, multi-tenancy, and context governance.
Read article
Continual learning is becoming boring, in a good way: it is the software development life cycle, applied to models. Model development requires systematic practices like versioning, APIs, and robust evaluation (Evals).
Read article
Context rot degrades agent reasoning as history piles up. Deep Agents mitigates this using offloading (saving large tool results to virtual file systems) and summarization (condensing context when token usage hits 85%).
Read article
The continual learning loop is unsolved. We must balance SFT (knowledge injection) and RL (behavior optimization) by carefully tuning the SFT/Distillation to RL ratio for optimal model performance.
Read article
To own your agent's intelligence, focus on three questions: where it fails, how to reproduce it, and how to fix it. This requires building robust 'worlds' and 'tasks' that mimic production.
Read article
Managed Deep Agents simplifies deploying mission-critical AI agents. It bundles the Deep Agents harness with managed LangSmith infrastructure, allowing developers to define, test, and deploy agents using the mda CLI.
Read article
LangSmith Engine v2 automates agent development, offering validated fixes, red teaming, and cross-trace analysis to improve AI agents before production.
Read article
Sydney Runkle details how to build a model router in the harness to reduce operational costs. By analyzing task traces and selecting models along the Pareto frontier, the router cut median cost per thread by 64%.
Read article
Web search can allow your agent to conduct external research and add that to information that you've already built in with your existing set of tools and skills. LangChain provides built-in search using parallel as part of the LangSmith platform.
Read article
LangChain outlines the shift to 'owning your intelligence' by building domain-specific agents. The platform provides Managed Deep Agents, Jev decision models, and LangSmith Engine v2 for robust governance and production readiness.
Read article
Learn how to build custom HTTP channels for Managed Deep Agents to handle inbound requests from external services like HubSpot or iMessage. We cover defining verifiers, parsers, and managing user context.
Read article
SmithTune provides a single, end-to-end workflow to convert agent traces into custom fine-tuned models. The process includes dataset creation, training (using providers like Fireworks or Baseten), and evaluation.
Read article
Learn how to use Schedules for Managed Deep Agents to automate recurring tasks. Configure cron jobs to generate weekly reports from Salesforce and deliver them automatically to Slack via LangSmith.
Read article
Jev is a System 1 model that makes fast, structured decisions without generating text. Learn how to integrate it into agent harnesses for safety classification and model routing, significantly improving efficiency and reliability.
Read article
LangSmith is LangChain's platform for the Agent Development Lifecycle (ADLC). It provides critical observability, allowing engineers to trace complex LLM agent interactions, debug tool calls, and monitor performance in production.
Read article
Learn how to build Patch, an agent that converts Slack messages into working GitHub PRs. This walkthrough details using Managed Deep Agents to integrate Slack, GitHub, and a sandbox with minimal code.
Read article
Jev, a new System 1 model from TypeSafe AI, is designed for fast, structured decision-making. It is up to 200x faster and 400x cheaper than LLMs for classification tasks, making it ideal for model routing and agent evaluation.
Read article
Learn how middleware extends Managed Deep Agents, enabling custom behaviors like PII redaction and tool call logging. See how to build custom middleware using `wrap_tool_call`.
Read article
Lyft rebuilt its customer support agent stack using a meta-agent architecture with LangGraph and LangSmith. This self-serve platform reduced agent build time from six months to two weeks, boosting resolution rate by 16%.
Read article