← Multimodal Sources 🕐 2 min read
Multimodal Sources

LangChain: What OpenClaw Got Right (And Why Enterprises Can't Have It)

> We see memory being represented as a set of files that gets pulled in in specific ways.

Show: Beyond the Pilot · Publisher: VentureBeat · Host: Matt Marshall, Sam Witteveen

Episode URL: https://traffic.megaphone.fm/UTEAU5140953619.mp3

Publish date: 2026-03-04
Duration: NAs
Default source credibility: HIGH — Named F500 practitioners on-record with production metrics. VentureBeat editorial vetting. Treat vendor-sponsored segments as MEDIUM.

  • Langchain’s Harrison Chase discusses the evolution of AI agents, emphasizing the importance of harnesses and observability for reliable agent deployment.
  • The discussion highlights the role of memory and context engineering in enhancing agent capabilities, with examples from enterprise use cases like Klarna and ServiceNow.
  • Langchain’s Langsmith platform is key for observability and evals, enabling enterprises to manage and improve AI agents effectively.

Extracted quotes

# Credibility Speaker Org Timestamp Topic Quote
1 HIGH Harrison Chase (CEO) Langchain 44:51 02-corporate-tools We see memory being represented as a set of files that gets pulled in in specific ways.
2 HIGH Harrison Chase (CEO) Langchain 52:44 02-corporate-tools We have our first like… no-code product, which is Langsmith Agent Builder, that is aimed kind of like at a similar audience.
3 HIGH Harrison Chase (CEO) Langchain 54:10 02-corporate-tools I think it’s in the observability, which powers evals, annotation queues, and debugging stuff.

Per-quote detail

1. Harrison Chase — Langchain (44:51)

We see memory being represented as a set of files that gets pulled in in specific ways.

  • Credibility: HIGH — Named exec at identifiable org with specific claim about memory representation.
  • Topic tag: 02-corporate-tools

2. Harrison Chase — Langchain (52:44)

We have our first like… no-code product, which is Langsmith Agent Builder, that is aimed kind of like at a similar audience.

  • Credibility: HIGH — Named exec at identifiable org with specific claim about new product.
  • Topic tag: 02-corporate-tools

3. Harrison Chase — Langchain (54:10)

I think it’s in the observability, which powers evals, annotation queues, and debugging stuff.

  • Credibility: HIGH — Named exec at identifiable org with specific claim about core value proposition.
  • Topic tag: 02-corporate-tools

Extracted 2026-04-15T00:20:40 via scripts/podcast_mine.py (MLX mlx-community/Qwen2.5-32B-Instruct-4bit).