LangChain Interrupt, Google & Microsoft AI, and the Future of Agent Tools
Tom and Cameron unpack the LangChain Interrupt Conference, new AI agent tools, and major updates from Google and Microsoft.
Now Playing
LangChain Interrupt, Google & Microsoft AI, and the Future of Agent Tools
Episode Summary
In this kickoff episode of The Build, Cameron Rohn and Tom Spencer dive deep into the LangChain Interrupt Conference, highlighting agent-first infrastructure, the rise of MCP protocols, and new evaluation standards. They break down LangChain's launch of LangGraph, Agent Inbox, and their prebuilt agent marketplace. Then they explore Google’s release of Gemini 2.5 Pro and Diffusion LLMs, and Microsoft's native MCP integration into Windows. The episode wraps with hands-on demos of Daytona, Cloudflare sandboxes, and speculative discussion about ambient agents, agent swarms, and just-in-time apps. Stay tuned for their live reaction to Claude 4’s surprise launch mid-recording.
Topics Covered
- LangChain Interrupt Conference themes and product updates
- LangGraph, LangSmith, Agent Inbox, and MCP integrations
- Agent Engineer as a new technical role
- Google’s Gemini 2.5 Pro, Diffusion LLMs, and creative tools
- Microsoft's MCP architecture and agent strategy
- Evals, open LLM judging, and agent observability
- Ambient agents and remote containerized execution
- Just-in-time software and AI-native runtime sandboxes
Key Takeaways
Takeaway 1
LangChain is driving adoption of agent-first architectures, with LangGraph and Agent Inbox leading the ecosystem.
Takeaway 2
Large enterprises like BlackRock and Nubank are deploying agents in high-risk environments, signaling production-readiness.
Takeaway 3
MCP and A2A (agent-to-agent) protocols are emerging as key components of the AI infrastructure layer.
Takeaway 4
Google and Microsoft are racing to embed agents into their OS and productivity ecosystems, with strong bets on creative tools and developer platforms.
Takeaway 5
Remote sandbox environments like Daytona and Cloudflare enable secure, scalable, just-in-time software execution for AI agents.
Featured Guests
No guests in this episode.
