Harrison Chase

Creator of LangChain and advocate for agent-centric AI development frameworks.

Mentions

Appeared in

Discussed in

Key positions and views

Access to a file system is a mandatory, non-negotiable requirement for building effective long-horizon agents in the current landscape.

The 'harness'—the framework, tools, and prompts around a model—is a critical performance lever, capable of producing dramatic improvements on benchmarks even with the same underlying model.

The source of truth for AI agents is a combination of code and execution traces, a fundamental shift from traditional software where code is the sole authority.

The most effective and reliable current applications for long-horizon agents are tasks that produce a 'first draft' for human review, such as coding, research, and report generation.

The ultimate strategic goal of the LangChain ecosystem is to create a self-improving 'AI, AI engineer' by building a closed feedback loop between agent execution, observability, and code modification.

Podcast consensus on Chase

Points of consensus

▶The 'harness' or scaffolding around a language model is a critical determinant of an agent's performance, sometimes more so than the model itself. This is supported by claims that tuning the harness improved a benchmark ranking from 30th to 5th, and that the popularity of tools like Anthropic's Claude Code is largely due to its harness.Apr 2026

▶Early agent frameworks like AutoGPT failed primarily because the underlying language models of the time were not capable enough to support their architecture.Apr 2026

▶Observability through detailed execution tracing is fundamentally more critical for debugging multi-step AI agents than for single-call LLM applications, as the context is dynamic and unpredictable.

▶Providing AI agents with tools to interact with a file system and code is currently a more effective and reliable strategy than giving them browser interaction tools.Apr–May 2026

Points of debate

▶Chase posits that the 'source of truth' for AI agent applications is a combination of code and execution traces, which contrasts with the traditional software development paradigm where code is the sole source of truth.Apr–May 2026

▶While acknowledging 'memory' as a key technological component for agents, Chase notes there are currently no industry standards for its structure or implementation, indicating a field in flux.Apr 2026

▶Chase argues against the current viability of purely asynchronous user interfaces for agents, advocating for a hybrid model that allows fluid switching between asynchronous management and synchronous chat, a point of debate in UX design for AI.

▶He observes that agent engineering teams are often staffed with more junior developers. This is a counter-intuitive staffing model given the complexity of the domain, contrasting with how senior talent is typically deployed on cutting-edge R&D.

Key themes

▶The Primacy of the 'Harness' Over the ModelApr 2026

Chase consistently argues that the framework, tools, and prompts surrounding an LLM (the 'harness') are as important, if not more so, than the underlying model's raw capability. This is evidenced by his claims about tuning LangChain's DeepAgents to dramatically improve benchmark scores and his analysis that much of Claude Code's popularity stems from its effective harness.

Investors and analysts should evaluate AI companies not just on their foundation models but on the quality and sophistication of their agentic frameworks and developer ecosystems, as these harnesses are a key differentiator for application performance.

▶Observability as the Cornerstone of AI DevelopmentApr–May 2026

He posits that for complex, multi-step AI agents, traditional debugging is insufficient. The dynamic and unpredictable nature of agent execution makes detailed tracing (as provided by LangSmith) the new source of truth, essential for both debugging and creating self-improving systems.

The MLOps market will likely see a major shift towards tools specializing in agent tracing, evaluation, and observability, as these capabilities are non-negotiable for building and maintaining reliable agentic applications at scale.

▶The 'AI, AI Engineer' VisionApr–May 2026

Chase's ultimate goal for the LangChain ecosystem is to automate the work of an AI engineer. This involves creating a closed-loop system where an agent (built with LangChain/DeepAgents) can use observability data (from LangSmith) to diagnose its own failures and rewrite its own code (using tools like Gemini Code Assist).

This vision suggests a future where the primary role of human developers shifts from writing code to designing, supervising, and refining these self-improving meta-systems, fundamentally altering the software development lifecycle.

▶Pragmatism in Long-Horizon Agent ApplicationsMay 2026

While ambitious, Chase maintains a practical view of current agent capabilities. He identifies coding and 'first draft' generation (e.g., research, reports) as the most viable applications due to agents' lack of perfect reliability, and he emphasizes foundational tools like file systems over more complex, less reliable ones like web browsers.

Near-term commercial success in the AI agent space will likely come from human-in-the-loop applications that augment professional workflows, rather than from fully autonomous systems that require near-perfect reliability.

Source episodes

Sentiment over time

Apr 2026

6 bullish, 1 bearish, 7 neutral(14 claims)

LangSmith provides an off-the-shelf evaluator that detects when an AI agent is s...

May 2026

1 bullish, 1 bearish, 3 neutral(5 claims)

A common and effective practice in building modern AI agents is to provide them ...

Changes over time

Early 2023

Notes the failure of early, popular agent projects like AutoGPT, attributing their decline to the insufficient reasoning capabilities of the underlying language models at the time.

Post-AutoGPT Era

Observes a shift in focus towards the 'harness' or scaffolding around models as a key performance driver, citing the popularity of Anthropic's Claude Code as an example where the framework is a major contributor to its success.

Current State

Identifies software development and coding as the primary domain where long-horizon agents are achieving significant success and adoption, framing their role as powerful assistants that produce a 'first draft' for human review.

Emerging Paradigm

Articulates a new development paradigm for AI, stating that the 'source of truth' for agent behavior is a combination of the code and the execution traces, a fundamental departure from traditional software engineering.

Future Vision

Outlines the strategic vision for LangChain and LangSmith to create a self-improving 'AI, AI engineer' by integrating agent frameworks with observability tools to form a closed feedback loop for autonomous debugging and code correction.

Suggested prompts

How does Harrison Chase's 'code + traces' paradigm for AI development challenge traditional DevOps and CI/CD practices? &nearr;Based on Chase's emphasis on the 'harness,' what are the key metrics for evaluating an agent framework independent of the underlying LLM? &nearr;What are the primary obstacles to achieving Chase's vision of a self-improving 'AI, AI engineer,' and how does the LangChain/LangSmith ecosystem aim to address them? &nearr;Chase dismisses browser tools as currently ineffective for agents; what specific model or tooling advancements would be required to change his assessment? &nearr;

Key concepts

AI Agents 2 ep LangSmith 2 ep LangChain 2 ep Agent Harness / Scaffolding 2 ep Observability & Tracing 2 ep Coding & Software Development 2 ep Foundation Models (Gemini, OpenAI, Anthropic) 2 ep Agent Memory 2 ep File System & Bash Tools 1 ep Long-Horizon Agents 1 ep

Notable quotes

“how we think about the next evolution of Lang Smith and Lang Chain is we're really creating this like AI, AI engineer.”

Harrison Chase · Automating the SDLC with LangChain, LangSmith, and Gemini

“for agents, it's a combination now of code and traces are where you can see the source of truth.”

Harrison Chase · Context Engineering Our Way to Long-Horizon Agents: LangChain’s Harrison Chase

“if you can find these framings where they run for a long period of time, but produce like a first draft of something, those to me are like the killer applications of long horizon agents right now.”

Harrison Chase · Context Engineering Our Way to Long-Horizon Agents: LangChain’s Harrison Chase