Stop re-explaining your company to AI agents
Pydantic and Dosu on company knowledge, agent memory, and reliable internal workflows. How to build internal agents that know how your company works: choosing a useful workflow, keeping their knowledge current, and using traces and evals to judge the results.
Pydantic and Dosu on company knowledge, agent memory, and reliable internal workflows.
Overview
- 60 min
- On demand replay
- Online event
- Free
Speakers
-
Michael Pfaffenberger Harness Engineer · Pydantic Harness Engineer at Pydantic. -
Douwe Maan Lead Developer, Pydantic AI · Pydantic Pydantic AI Lead Engineer
-
Devin Stein Founder & CEO · Dosu Founder and CEO of Dosu, which builds AI agents that answer engineering questions and keep project knowledge current. -
Taylor Dolezal Head of Open Source · Dosu Head of Open Source at Dosu, working on how maintainers keep project knowledge answerable. Previously Head of Ecosystem at the Cloud Native Computing Foundation. -
Laís Carvalho Developer Relations, host · Pydantic Growth marketing & Developer Relations at Pydantic. Passionate about Python, open source, and building great developer experiences.
About this recording
Internal AI agents need the knowledge of the company they work inside. Whether they answer employee questions, investigate operational issues, or prepare finance reports, they need the right tools and current company knowledge. How do you provide that knowledge, keep it up to date, and measure whether it improves their work?
In this fireside chat, Pydantic and Dosu talk through building internal agents for everyday work. Dosu builds agents with Pydantic AI and traces their runs with Pydantic Logfire. Drawing on that experience, the conversation covers how teams build agent workflows, investigate failures, and maintain the company knowledge their agents use.
Devin Stein, Taylor Dolezal, Michael Pfaffenberger, and Douwe Maan discuss how to choose an internal workflow, where to put a human review step, and how to use traces to find the causes of failures. They also get into what agents should remember between tasks and how to test whether that memory improves the results.
If you develop agents for your organization, or lead a team adopting them, the recording covers what to build, which knowledge to provide, and how to evaluate the work.
What you'll learn
- How to pick an internal workflow worth handing to an agent, from employee questions to operational investigations to finance reporting
- Ways to give an agent current company knowledge, and keep it current as the company changes
- Where a person should review the agent's work, and how to define that boundary
- Reading Logfire traces to find the cause of a failed agent run
- What an agent should remember between tasks, and how to test whether that memory improves results
Watch the recording
Tell us where to send Pydantic updates and unlock instant access to this replay.
You're all set
Your request was received. You can close this page.
We couldn't open the recording
Please go back and try again in a moment.
Explore Logfire
Explore our open source packages
