Managing multiple Hermes workflows exposed unexpected challenges

Running multiple Hermes workflows sounded simple… until it wasn’t.

At Agent37, we’ve been using Hermes for real work:

• Coding tasks
• Research workflows
• Long-running automation jobs
• Background tasks throughout the day

Initially, the main challenge was simply keeping Hermes running smoothly. Once setup and hosting were stable, a new problem emerged: managing the workflows themselves became increasingly complex.

Key issues included:

• Multiple terminal sessions open simultaneously
• Unfinished jobs across projects
• Tasks spread across different workflows
• Limited visibility into:
  • What had completed
  • What was stuck
    - What still required review

The agents themselves worked as expected. The real bottleneck was the operational and supervisory layer around them, which made managing multiple workflows inefficient and frustrating.

This experience highlighted the need for a dedicated solution, which led us to build an internal Mission Control system for Hermes workflows.

I’m curious have others faced similar challenges with long-running AI workflows? How do you stay organized across multiple sessions?