if you have a team of grok bots, here's something you need to try
i gave my chief of staff bot an assignment:
help every other bot on my team get better at its job
the instructions were pretty simple
for each bot, i asked it to:
1. find 3 books related to its role
2. research the ideas from those books
3. save the useful knowledge as a skill
4. decide how it could apply that knowledge to its work
5. report back to the chief of staff with what it learned and what it plans to do differently
so instead of me researching books for every role and writing instructions for every bot myself
i gave the job to the person... well, bot, that already coordinates the team haha
the part i find interesting is saving the knowledge as a skill
i don't just want a summary sitting in a chat that gets forgotten.
i want the bot to have something it can refer back to when it's actually doing its job
for example, if a bot helps me write video scripts, it could research storytelling and hooks, save the useful principles, and use them the next time i ask for a script
then the chief of staff gets a report on what each bot found and how it plans to use it
i'm still testing this, so i'm not going to pretend my bots suddenly became experts or that every book was read cover to cover. i'll have to check the sources and see if their work actually gets better
but as an experiment, i like the idea of giving each bot a way to improve its own playbook instead of starting from scratch every time
basically sent my entire bot team back to school haha
if you try it, i'd be curious what books your bots pick
@yeemio exactly, that's the part i'm most curious about too
a bot saying it learned something is one thing, actually seeing it do a better job afterwards is another haha
if you have a team of grok bots, here's something you need to try
i gave my chief of staff bot an assignment:
help every other bot on my team get better at its job
the instructions were pretty simple
for each bot, i asked it to:
1. find 3 books related to its role
2. research the ideas from those books
3. save the useful knowledge as a skill
4. decide how it could apply that knowledge to its work
5. report back to the chief of staff with what it learned and what it plans to do differently
so instead of me researching books for every role and writing instructions for every bot myself
i gave the job to the person... well, bot, that already coordinates the team haha
the part i find interesting is saving the knowledge as a skill
i don't just want a summary sitting in a chat that gets forgotten.
i want the bot to have something it can refer back to when it's actually doing its job
for example, if a bot helps me write video scripts, it could research storytelling and hooks, save the useful principles, and use them the next time i ask for a script
then the chief of staff gets a report on what each bot found and how it plans to use it
i'm still testing this, so i'm not going to pretend my bots suddenly became experts or that every book was read cover to cover. i'll have to check the sources and see if their work actually gets better
but as an experiment, i like the idea of giving each bot a way to improve its own playbook instead of starting from scratch every time
basically sent my entire bot team back to school haha
if you try it, i'd be curious what books your bots pick
@ArgileNoland right there with you man haha
trying to keep my AI costs as low as possible until my projects start bringing in enough to justify spending more
@matt_teeixeira i wonder if we'll end up with websites built primarily for agents rather than humans
the whole business model around traffic could change
@samongaro_ i try to find people already dealing with the problem and talk to them before building anything
even better if i can get someone to commit to paying for a simple version
still learning this myself though haha
@TonyChanBuilds that's interesting, so almost half your followers came from one post haha
really shows how much a single post can change things, even when you can't replicate it right away
@b4dgerr a whole year building before finding out he had no actual plan to sell it is brutal
makes me think co-founders should probably validate their partnership with a small project before committing to something big