Session Q&A Recap
Thanks to everyone who joined live and asked thoughtful questions during the Agent demo session! Here’s what we covered:
Agent Capabilities & Use Cases
Q: Can agents interact with databases?
A: Yes! Agents can read from or write to databases by invoking workflows, tools, or queries. For example, they can update records in a table after extracting structured data.
Q: Can I trigger an agent on a schedule?
A: Yes. You can use a scheduled Workflow to trigger an agent automatically—no manual chat required.
Q: Can agents respond to email in a thread?
A: Yes! You can trigger agents from email, and if the right context is preserved, they can respond and continue the conversation across multiple threads.
Q: Can agents automate data extraction from receipts or emails?
A: Absolutely. You can pass in uploaded documents (like receipts or emails), then use an AI action to extract the info. Pair with a confirmation step before pushing to your system.
Q: Can I automate browser sessions?
A: Kind of. There’s no built-in tool yet for full browser automation, but you could integrate with something like Playwright via an MCP server if needed.
Q: Can I use agents for logistics tasks like vehicle route optimization?
A: Yes! You can build agents that combine APIs, vector search, and workflows to handle custom logistics scenarios. Check out this example.
Q: Can different agents or tools use different authentication?
A: Yes. Resources set up in Retool (e.g., APIs, DBs) carry their own credentials. We're working on more robust multi-key management for AI resources specifically.
Monitoring, Logs & Limits
Q: Are there limits on logs or storage?
A: Logging limits depend on whether you’re on cloud or self-hosted Retool. There’s no hard cap, but storage practices may differ.
Q: Can we access logs programmatically?
A: You can view logs in the monitoring dashboard and access run context from apps or workflows. Direct programmatic access depends on your deployment.
Q: Is there a cost for web search tools in agents?
A: Not today, but usage limits apply. Retool currently uses Tavily Web Search. Bring your own API key for full access.
Video Agent & Demo Tools
Q: Can I see the full agent structure from the demo?
A: Yes! We’ve added the video and JSON to the resources folder.
Q: What tools powered the video production agent?
A: It combined web search, a voiceover generator (Eleven Labs), an image-to-frame tool (the GPT-1-image model from OpenAI), and Shotstack for stitching the final video. All were triggered through Retool workflows.
Miscellaneous
Q: Do I need to register again for the rescheduled Monday session?
A: Nope! If you registered for Thursday, you’re already signed up for Monday. It won’t let us change the event on your calendar, but you can do that yourself 
Q: Does Retool have a startup program?
A: Yes—learn more here. Startups under 10 years old and under $10M in funding are eligible.
Q: Is Agents still in beta, and is it available globally?
A: Yes, still in beta—but it’s available anywhere Retool runs (cloud and self-hosted, including outside the U.S.).
Q: Can I use OpenAI’s Deep Research API in an agent?
A: Not natively yet, but you can set up a custom provider and integrate it with your workflows. Submit a feature request if you'd like deeper support.
💬 What did you think of this session?
Drop a comment — or vote in the quick poll below:
Loved it — more like this please!
It was solid, but I had questions.
Not quite what I expected.