Grant Scout
Find and compare relevant funding opportunities.
Give AI a goal—and the tools to accomplish it.
Build an agent that can reason through a task, use external tools, work across multiple steps, and return a useful result with human approval where it matters.
Basic familiarity with JavaScript is helpful, but agent experience is not required.
What you’ll build
A small AI agent that plans a task, selects approved tools, records its work, and pauses before consequential actions.
A working example
Try: “Find relevant open grants for a neighborhood technology program and prepare a shortlist for my review.”
The agent searches approved sources, checks eligibility, compares deadlines, and asks for approval before saving a final shortlist.
4 eligible grants found
Review required before export
What you can make
Use the project for your community, organization, business, research, or idea.
Find and compare relevant funding opportunities.
Collect context and prepare a concise briefing.
Track updates from a defined set of sources.
Complete a repeatable workflow with approval gates.
What you’ll learn
Finish with a working project and a practical understanding of the choices behind it.
Understand how agents differ from chatbots
Define a clear goal and stopping condition
Give an agent a limited set of tools
Carry context through a multi-step task
Validate tool inputs and structured outputs
Add human approval before consequential actions
Inspect traces, failures, cost, and reliability
What we’ll use
You do not need to know these tools beforehand. The guide introduces each one when you encounter it.
Coordinate reasoning and tool calls
Plan the task and interpret results
Let the agent retrieve or transform information
Keep a person in control of important actions
How the guide works
Give the agent one bounded goal and clear limits.
Add tools and test how the agent selects them.
Add review points, inspect traces, and improve reliability.
Ready to build?
Part of the AFAI AI Toolkit · Practical, open resources for everyone