Ask HN: What do you do when your AI agents are working?

There are basically two answers to this.

One option is to orchestrate more agents across different repositories or projects at the same time, so there’s essentially less downtime while you’re waiting between tasks. But I really dislike that kind of constant context switching - it makes me feel a bit like Lucille Ball in that 1950s episode where she's wrapping chocolates [1].

I prefer to focus on a single project at a time. When I do hand something off to an LLM, I have three go‑to activities that I alternate between. I'll try to do some physical exercise (jumping jacks, pull-ups and push-ups), grab my violin/piano since both are less than a meter from my desk, or I'll study Japanese.

[1] - https://youtu.be/AnHiAWlrYQc

I generally try to get a few parallel workstreams going, and try to cap concurrent tasks based on effort of build/review. For me the bigger bottleneck is when a number of agents finish simultaneously and all require big reviews requiring thoughtful project context.

So i try to stagger the big rocks and chip off smaller ones in parallel. Also being able to complete the smaller tasks and move on makes me feel less brain mush from parallelizing.

I use a project decision layer too to provide guidance across sessions. This makes me feel more confident in their work staying aligned with project direction.

Wonder how others determine when agents are running normally or when they need some human judgement.

In a work environment (at work), I often use Claude Code to fulfill work I would otherwise normally do with my steer and guidance. What I then do is go back to discussions, answering questions, reviewing others work, etc, until Claude has something for me to review.

This allows me to basically do "asynchronous" work and keep-up with the increasingly complex work environment where there are lots of things going on, sometimes too many things, all at once.

Working on multiple tasks/projects in parallel.

The crux is to know where your attention is needed and quickly get the full context of the task when you come back.

That’s why I built Codehydra https://github.com/stefanhoelzl/codehydra

This is a very common situation. By inserting another task during the waiting time, that waiting time is eliminated.

It is important to make good use of both manual work and AI.

Drink coffee

I mostly read books. Or I collect programming keywords to use AI better

[flagged]