A practical validation problem for AI agent workflows
An AI agent is being used to build a graph made of connected workflow steps. Each node is a JSON structure, and a node can take inputs such as a URL, an email sender, or an email receiver. Each node points to the next node, so later steps can use the output from earlier steps when needed.
The workflow may also contain loops. The graph can be structurally correct, but the LLM often creates the wrong s between nodes. It does not reliably know how one node should use the input or output from the previous node, and it also struggles with loops.
A step after graph creation is being considered. One option is to run the workflow with , but that could require nearly 600 accounts across services such as Gmail, Slack, and Microsoft; another option is to treat it as a problem.
Key points
- An AI agent is generating connected workflow graphs.
- Each node is a JSON structure with inputs such as URLs and email fields.
- The LLM often gets s between nodes wrong, even when the graph shape is valid.
- Loops make the problem harder.
- Running with real could require nearly 600 Gmail, Slack, Microsoft, and similar accounts.