Uploaded August 2026 | Updated November 2026, 2 days ago
Ravi Chandu Ummadisetti, Director of Agentic AI and Product Research at Toyota North America, explains how his team built R&D GPT, an internal agent that searches Toyota's research, paint, and manufacturing data to answer engineering questions. He breaks down how Deep Agents' single command and harness pull in skills from across Toyota's research, branding, and supply chain teams, and why Polly and Insights inside LangSmith are the tools he relies on most to understand what his agents are actually doing in production.
Chapters:
0:00 Meet Toyota's Director of Agentic AI and Product Research
0:31 What makes Deep Agents different: one command, one harness
0:50 Pulling paint corrosion and research data into a single agent
1:07 How engineers query Toyota's research in plain language
1:24 Turning Toyota's institutional knowledge into reusable agent skills
1:44 Why Polly and Insights are Ravi's favorite parts of LangSmith
1:56 Watching what agents actually do in production
Extra resources:
→ LangSmith: langchain.com/langsmith
→ Deep Agents: langchain.com/deep-agents
Ravi Chandu Ummadisetti, Director of Agentic AI and Product Research at Toyota North America, explains how his team built R&D GPT, an internal agent that searches Toyota's research, paint, and manufacturing data to answer engineering questions. He breaks down how Deep Agents' single command and harness pull in skills from across Toyota's research, branding, and supply chain teams, and why Polly and Insights inside LangSmith are the tools he relies on most to understand what his agents are actually doing in production.
Chapters:
0:00 Meet Toyota's Director of Agentic AI and Product Research
0:31 What makes Deep Agents different: one command, one harness
0:50 Pulling paint corrosion and research data into a single agent
1:07 How engineers query Toyota's research in plain language
1:24 Turning Toyota's institutional knowledge into reusable agent skills
1:44 Why Polly and Insights are Ravi's favorite parts of LangSmith
1:56 Watching what agents actually do in production
Extra resources:
→ LangSmith: langchain.com/langsmith
→ Deep Agents: langchain.com/deep-agents










