Hi everyone. I’m fairly new to SuperAGI and have been exploring it over the past few weeks, primarily to see how it could handle some basic HR-related workflows. I wanted to share my experience building a simple agent to parse and triage inbound emails—something like sorting candidate inquiries or internal support requests.
I started with a clear but narrow goal: an agent that reads a simulated email, extracts key entities like the sender's name and the core request, and then decides on a priority level and routing. I used the SuperAGI GUI, which felt intuitive for a beginner. The key was in carefully designing the initial prompt within the agent configuration. I described the exact format I wanted the output in (like a JSON structure with fields for priority, category, and next action) and provided a few clear examples of input emails and the desired parsed output. This step was crucial for getting consistent results.
The tools I attached were straightforward: just the Read File and Write File tools for this test. I stored a sample email in a .txt file for the agent to read. Running the agent was a bit slow on my machine, but it worked reliably. The main challenge was tweaking the prompt to handle edge cases, like vague subject lines or multiple requests in one email. After a few iterations, the agent became quite good at its specific task.
I can see this being a foundation for more complex onboarding or employee query systems. Has anyone else built something similar for handling unstructured text? I’d be curious to learn about your approaches, especially if you’ve integrated external tools or APIs for a more live workflow. Thank you for any insights you can share.