Uploaded July 2026 | Updated November 2026, 1 day ago
Ben Sklar, Director of Product Management at Salesforce, explains how the team behind Agentforce Vibes, Salesforce's enterprise vibe coding product, moved from teams building their own one-off MCP tools to a shared evaluation standard with LangChain and LangSmith.
In this video, Ben walks through how LangSmith lets them:
✅ Run evals at scale before code reaches customers
✅ Standardize scoring across instruction following, coherence, factuality, and deployability
✅ Support the milestone of over 100 million lines of accepted code
Chapters:
0:00 The problem with every team building its own MCP tools
0:22 Making sure agent output does what you expect
0:35 Finding LangChain and LangSmith
0:53 Running evals at scale before code reaches customers
1:19 Setting one scoring standard across every team
1:40 100 million lines of accepted code
Get started with LangSmith: langchain.com/langsmith-platform
Ben Sklar, Director of Product Management at Salesforce, explains how the team behind Agentforce Vibes, Salesforce's enterprise vibe coding product, moved from teams building their own one-off MCP tools to a shared evaluation standard with LangChain and LangSmith.
In this video, Ben walks through how LangSmith lets them:
✅ Run evals at scale before code reaches customers
✅ Standardize scoring across instruction following, coherence, factuality, and deployability
✅ Support the milestone of over 100 million lines of accepted code
Chapters:
0:00 The problem with every team building its own MCP tools
0:22 Making sure agent output does what you expect
0:35 Finding LangChain and LangSmith
0:53 Running evals at scale before code reaches customers
1:19 Setting one scoring standard across every team
1:40 100 million lines of accepted code
Get started with LangSmith: langchain.com/langsmith-platform










