@brooksjordan: This is 100% my experience. There is agent automation, which is powerful, but there is also human-agent collaboration, …

This is 100% my experience. There is agent automation, which is powerful, but there is also human-agent collaboration, if you want the work to be great:

From @every

“You might already see, in the midst of all of this automation, where the humans come in. In every example, the agent needs a human in order for the work to, well, work.

“Someone has to point it at the right thing, decide whether the output is good, catch the places where it is wrong, and turn the result into a real-life decision or process.”

I agree!

Say I have four to five agents doing substantial pieces of work and often multiple sub-agents working for those primary agents

A human is necessary because knowledge isn’t understanding, which is priceless

I know in my heart of hearts that if I hadn’t asked critical questions at the right point or applied higher understanding to an architecture or caught details about stale thinking or code, I would start to get slop

The question is can you automate even aspects of that human-agent collaboration? This could be done with better planning and priorities, having agents review another’s work via MCP, putting eyes on it at different points in the process, using better models and harnesses, etc.

It will evolve but it’s amazingly helpful to have @danshipper and the Every team call out how their humans are skillfully working with agents today when understanding and decisions are required

I’m the guy who needed to read Jason’s post today:

“It’s hard to live with something that isn’t quite right yet – especially when it’s your job to get it right.”