GenAI PMHarrison Chase
Founder and AI developer advocate associated with agent tooling and workflows. Here he discusses defining agents with markdown and JSON files for streamlined development.
Key Highlights
- Harrison Chase is a key public voice shaping how teams build, evaluate, and operationalize AI agents.
- His recent updates emphasize agent middleware, tracing, deployment workflows, and continual improvement loops.
- He frames LangSmith and agent harnesses as foundational infrastructure for production-grade agent systems.
- He has highlighted long-term memory as an important next frontier for practical AI agents.
- For AI PMs, his work offers concrete patterns for moving from agent prototypes to reliable products.
Harrison Chase
Overview
Harrison Chase is a prominent builder and public voice in the AI agents ecosystem, best known as the founder of LangChain and a leading advocate for practical agent development workflows. In recent newsletter coverage, he is repeatedly associated with the infrastructure layer around agents: agent harnesses, middleware, tracing, evaluation, deployment, memory, and workflow definition through markdown and JSON files. His commentary and product announcements consistently focus on making agents easier to build, customize, observe, and improve in production.For AI Product Managers, Harrison Chase matters because his work sits at the intersection of agent tooling and operationalization. Across mentions, he frames agents not as one-off demos but as systems that require middleware, evaluation readiness, long-term memory, deployment hooks, and iterative improvement loops. That makes him a useful signal for where the agent stack is heading—especially for PMs deciding how to standardize internal agent architectures, choose observability platforms, and move from prototypes to reliable products.
Key Developments
- 2026-03-24: Highlighted webhook support in LangSmith Deployments, enabling teams to trigger Slack notifications or downstream actions when long-running agent runs complete.
- 2026-03-27: Shared a blog post on a rigorous evaluation framework for real AI agents, covering scoring rubrics, simulations, and benchmark analysis beyond simple prompt testing.
- 2026-03-28: Pointed to Vic’s LangChain Agent Evaluation Readiness Checklist as a practical guide for taking agents into production.
- 2026-04-01: Explained how to run a continual agent improvement loop with LangSmith using trace-centered iteration.
- 2026-04-03: Showcased how @vishsuresh_ built an automated feedback loop for a GTM agent, emphasizing repeatable improvement mechanisms.
- 2026-04-07: Highlighted LangChain’s community middleware page and positioned agent middleware as a key mechanism for tailoring agent harnesses to specific use cases.
- 2026-04-08: Announced that LangSmith Fleet integrates with Arcade.dev, unlocking enterprise-grade access to 8,000+ tools and enabling no-code Claude Cowork/OpenClaw-style agents.
- 2026-04-08: Unveiled LangSmith’s tracing and evaluation platform, emphasizing diagnosis and optimization of agent behavior in real-world conditions.
- 2026-04-10: Noted that community middleware such as langchain-task-steering is emerging to customize agents and deepagents, and invited contributors to coordinate via Sydney.
- 2026-04-11: Likened agent harnesses to Spark and positioned LangSmith as the Databricks of agent abstractions, signaling a maturing infrastructure layer for agent development.
- 2026-04-11: Confirmed that long-term memory in AI agents will be a major focus going forward.
Relevance to AI PMs
1. Framework for productionizing agents Harrison Chase’s updates consistently emphasize evaluation, tracing, deployment, and feedback loops. For PMs, this is a practical reminder to define success metrics, observability, and post-launch iteration processes before scaling any agent product.2. Signals on the emerging agent stack
His references to agent harnesses, middleware, LangSmith, deepagents, and long-term memory help PMs understand which layers of the stack are becoming standardized. This is useful when deciding whether to build custom orchestration in-house or adopt ecosystem tools.
3. Operational patterns, not just model features
Chase’s focus on markdown/JSON-defined agents, middleware extensions, and trace-based improvement loops suggests that product differentiation may increasingly come from workflow design and reliability engineering rather than model selection alone.
Related
- LangChain: Harrison Chase is most closely associated with LangChain, the broader ecosystem for building LLM and agent applications.
- LangSmith: Frequently connected to his work on tracing, evaluation, deployments, and agent improvement loops; central to his production-oriented messaging.
- LangGraph: Relevant as part of the structured agent workflow and orchestration layer around LangChain-based systems.
- deepagents / deep-agents / deepagents-cli: Connected through discussion of customizable agents and middleware patterns.
- agent-harnesses: A recurring concept in his framing of reusable abstractions for agent building.
- agent-middleware / langchain-task-steering: Illustrate the extension layer he highlights for adapting agents to specific business use cases.
- Arcade.dev: Connected through LangSmith Fleet integration for large-scale tool access.
- Sydney: Mentioned as a contact point for middleware contributions in the ecosystem.
- Vic: Referenced through the LangChain Agent Evaluation Readiness Checklist that Chase endorsed for production readiness.
- memory: A strategic theme he explicitly called out as a major future focus for AI agents.
- traces / code / agents / markdownjson-files: These themes reflect his emphasis on observable, configurable, developer-friendly agent workflows.
Newsletter Mentions (26)
“Harrison Chase likens agent harnesses to Spark and positions LangSmith as the Databricks of agent abstractions, quoting @bllchmbrs’ analogy of them as stable building blocks.”
#21 𝕏 Harrison Chase likens agent harnesses to Spark and positions LangSmith as the Databricks of agent abstractions, quoting @bllchmbrs’ analogy of them as stable building blocks. #22 𝕏 Harrison Chase confirms that long-term memory in AI agents will be a major focus going forward.
“#25 𝕏 Harrison Chase notes that community middleware—like “langchain-task-steering”—is popping up for customizing agents and deepagents, and invites anyone with middleware to contribute by reaching out to Sydney.”
#25 𝕏 Harrison Chase notes that community middleware—like “langchain-task-steering”—is popping up for customizing agents and deepagents, and invites anyone with middleware to contribute by reaching out to Sydney.
“Harrison Chase notes that community middleware—like “langchain-task-steering”—is popping up for customizing agents and deepagents, and invites anyone with middleware to contribute by reaching out to Sydney.”
#25 𝕏 Harrison Chase notes that community middleware—like “langchain-task-steering”—is popping up for customizing agents and deepagents, and invites anyone with middleware to contribute by reaching out to Sydney.
“Harrison Chase announced that LangSmith Fleet now integrates with Arcade.dev, offering enterprise-grade access to 8,000+ tools and enabling you to build no-code Claude Cowork/OpenClaw–style agents in minutes.”
#8 𝕏 Harrison Chase announced that LangSmith Fleet now integrates with Arcade.dev, offering enterprise-grade access to 8,000+ tools and enabling you to build no-code Claude Cowork/OpenClaw–style agents in minutes. #9 𝕏 Harrison Chase unveils LangSmith’s tracing and evaluation platform—spotlighted on new SF & NYC billboards—to help teams track, diagnose, and optimize agent behavior in real-world conditions.
“#6 𝕏 Harrison Chase highlights LangChain’s new community middleware page, showcasing agent middleware as a powerful way to tailor agent harnesses to specific use cases.”
#6 𝕏 Harrison Chase highlights LangChain’s new community middleware page, showcasing agent middleware as a powerful way to tailor agent harnesses to specific use cases. He’s inviting developers to share what they’re building with these middleware integrations.
“Harrison Chase showcases how @vishsuresh_ built an automated feedback loop for their GTM agent, with step-by-step implementation details available in the linked blog post.”
#11 𝕏 Harrison Chase showcases how @vishsuresh_ built an automated feedback loop for their GTM agent, with step-by-step implementation details available in the linked blog post. #12 𝕏 Santiago Pika now lets you send Google Meet invites to agents like OpenClaw or Claude so they can join live video calls, enabling true “face-to-face” AI interactions.
“Harrison Chase explains how to power a continual agent improvement loop with Langsmith, using trace-centered iteration from LangChain’s “agent improvement loop” guide.”
𝕏 Harrison Chase explains how to power a continual agent improvement loop with Langsmith, using trace-centered iteration from LangChain’s “agent improvement loop” guide.
“#4 𝕏 Harrison Chase points to Vic’s LangChain Agent Evaluation Readiness Checklist as a go-to, step-by-step guide for taking AI agents into production.”
#4 𝕏 Harrison Chase points to Vic’s LangChain Agent Evaluation Readiness Checklist as a go-to, step-by-step guide for taking AI agents into production.
“Harrison Chase shares a blog post detailing his team’s rigorous evaluation framework for real AI agents, not just simple LLM prompts.”
#18 𝕏 Harrison Chase shares a blog post detailing his team’s rigorous evaluation framework for real AI agents, not just simple LLM prompts. It walks through scoring rubrics, simulation setups, and benchmark analyses to quantify agent capabilities.
“Harrison Chase adds webhook support to LangSmith Deployments so you can send Slack pings or trigger other actions automatically when a long-running agent run finishes.”
#8 𝕏 Harrison Chase adds webhook support to LangSmith Deployments so you can send Slack pings or trigger other actions automatically when a long-running agent run finishes.
Related
Anthropic's coding-focused agentic tool for building and automating software workflows. In this newsletter it is discussed as being integrated with Vercel AI Gateway and as a Chrome extension for browser automation.
An open-source digital assistant built on Claude Code that can manage emails, transcribe audio, negotiate purchases, and automate tasks via skills and hooks.
A company building developer tools and frameworks for LLM applications. It is implied by the mention of Harrison Chase and a product progress bar/pricing discussion.
Autonomous or semi-autonomous systems used here in sales and coding workflows. The newsletter highlights their role in replacing human SDR tasks and orchestrating complex tasks.
A platform for evaluating and debugging LLM and agent pipelines. In this issue it is cited for a redesigned experiment comparison view for side-by-side benchmarking.
A Claude-based workflow used here to identify key skills for the AI era.
A modular layer that adds tools, guardrails, and custom instructions to AI agents. It is described as a composable harness for production agent systems.
A component or pattern used in LangSmith Agent Builder to support more capable agent workflows.
Community middleware example for customizing agent behavior and steering tasks in agent frameworks.
Stay updated on Harrison Chase
Get curated AI PM insights delivered daily — covering this and 1,000+ other sources.
Subscribe Free