Carl Yeh

So what you're doing is every single time you complete a task, in this case, this tab, there's a task that they do and they review 30, 40, 50 PDFs at a time, extract information and rename, do a rename and then move into a different folder.

Every single time it does that task, I have in here a slash command called retrospective.

And so when you hit retrospective,

It does three things.

One is it reviews your entire session, like your conversation for that session.

Then it takes the learnings and failures.

And then it updates the learning and the skill for each agent, subagent here.

So every single time we complete a task, I just hit retrospective.

So everything that we learn from that task, any edge cases, is automatically populated into the learnings and skill.

And every single time the agent executes, it goes into its learnings and its skill.

So hence, continuous learning every single time it's used.

So that's pretty much it.

There are some use cases.

there's some use cases that, I mean, sorry, edge cases that we haven't seen before.

And so it's like, oh, in one instance, like if the property code has these extra three letters, what do we do?

I was like, oh, we do this, this, this, but that gets written up in the retrospective and then automatically populates into learning and skill.

So the next time you don't have to remember that it's already part of the learning and skill that,

It's written over.

So, yes, essentially you are self-improving the agent, especially on the skill side.

So you do not have to reprompt any edge case or any specifics outside of like any major overhaul you need to do with this.