
Wassup Bot maestro,
The personal-agent party is getting crowded. Meta just announced Muse for Small Business, with connections to Shopify and Klaviyo. OpenAI has Dots. And Grok Bots just gave us Team Bots.
My take: pick something that fits how you actually work.
Aside from Grok Bot, Muse has been the most delightful experience I've had. Watching your little agent work in the background is surprisingly fun. For a solo operator, especially someone already working across Meta's apps, I think it's a solid place to start.
I've only had a brief go with Dots. It felt buggy and laggy in my testing, so this is an early impression. I'll keep playing with it.
For a small or medium-sized team, Grok Bots is my pick right now. The shared setup is what interests me: give people the same files, skills and instructions, then let them build on each other's work.
Because you know how this usually goes. Someone learns something useful in a client call. Someone else has the research in a Google Doc. The decision lands in a Slack DM. Good luck finding all of that before the next meeting.
Here's a job I'll be testing at Z***** (new agency in stealth): a project-manager Bot that collects approved research, decisions and next steps into our existing Notion project record. A teammate could use that research for an outline or creative brief, check the sources, and add what they learned back to the record.
That takes an explicit handoff. Private chats don't become shared team knowledge just because everyone uses the same Bot. We haven't proven this workflow yet, but I want to try it because helping people learn from each other's work gets me jacked up!
Now, back to the Grok Bot upgrades and alpha.
On Monday September 28, xAI announced Team Bots in public beta for Teams and Enterprise plans. A team can share a Bot’s files, skills and setup while each teammate has a private chat. The Bot can also join Slack with its own handle.
How the team plan works
The subscription gives people access; a Team Bot is the shared worker they use. Its owner sets up the files, skills and connections, tries it on real tasks, then publishes it to the team. Teammates find it under New chat → Team Bots. The owner maintains the shared setup, so everyone uses the same version. Setup guide.
Private chats stay separate; conversations in a shared Slack channel are visible there. Connected accounts use each person’s sign-in where required. A shared service key gives the Bot access for everyone using it, so choose its permissions carefully.
What it costs
Cursor’s team pricing lists:
Standard at $40 per person per month
Premium at $120 per person per month
Subscriptions include weekly Bot usage. Eligible accounts can add on-demand usage charged by model and token cost, so the seat price isn’t a promise of unlimited work. Check your account’s allowance and on-demand settings before choosing a routine.
When a teammate asks a Team Bot to work, it uses that teammate’s Bot allowance. A routine uses its creator’s allowance; Slack workflow messages without a linked teammate use the owner’s. Usage rules.
Where it could help across your org
These are suggested starting jobs, not results from testing the new release:
Marketing: give it the approved brand guide, product facts and offer terms. Ask it to flag unsupported claims or off-brand wording in a campaign draft, with the source beside each flag.
Customer experience: give it current returns and shipping policies. Have it draft a response for an agent and identify cases that need a person’s decision. Keep refunds and customer sends for human approval.
Merchandising and operations: give it a launch checklist and a dated product-data export. Ask for missing images, mismatched offer terms and unresolved handoffs, with links to the evidence. Review the list before changing the store.
Agency or account teams: use client-approved files and connections with access limited to the relevant people. Use approved notes and the task board to prepare a handoff showing what was promised, who owns it and what is still unconfirmed. Limit access to the relevant team.
Start with one shared job and one person responsible for its setup. Team Bot routines are personal: they report to the person who creates them. If the team needs the output, have the routine post to an agreed shared destination.
Now, a few small things to try with your Bots this week.

3 Tips & Tricks
01. Make your Bot explain its moves
Learning a new tool while trying to get actual work done? Ask your Bot to explain unfamiliar terms and what it’s doing as you go.
Michael Fenech shared a prompt for this. Try it on a real job, like putting a launch brief together:
As we work on [job], explain unfamiliar features or terms the first time they come up. Before taking an action, tell me what you’re about to do and why. Keep it brief, and ask me before changing or sending anything.
If it explains something incorrectly, check the product’s docs and give it the correction. You’re giving it an instruction to explain the work as you go.
02. Pause the routines you no longer need
That daily report you set up for a launch can keep running long after the launch is over. Open your routines and check which ones you still want.
Texas Twins Dad’s reminder: review the repetitive jobs. Pause the stale ones. For a routine you’re keeping, make the reporting rule specific:
For this routine, report when [condition worth my attention]. Include the source and what changed. If you can’t reach the source, tell me the check failed. Otherwise, keep the update brief.
Check the saved schedule after you change it. Asking a Bot to stay quiet doesn’t stop the routine from running or using its allowance. Pause the job itself when you don’t need it.
03. Give each Bot a clear job in your email workflow
Jesse Streur shared his Grok Bot email setup with different Bots handling different roles. Give each Bot a clear job, then make the handoffs explicit.
Start with the campaign you’re working on. Give an analysis Bot the approved performance report and ask what’s worth testing. Pass the reviewed findings to a planning Bot for the campaign outline. Then give the approved outline, product facts and offer terms to your drafting Bot.
Your job is [analysis, planning or drafting]. Use [approved inputs]. Return [specific output] for the next Bot. Link the sources, flag anything missing, and wait for my approval before passing it along or changing anything.
Make each handoff explicit. Attach the approved output or save it somewhere the next Bot can read. You review the campaign and send it. Start with a single email so you can see where the handoff needs work.

2 Bot Templates
These recommendations describe the marketplace listings and suggested first runs. The templates haven’t been imported or run as part of this issue’s preparation.
01. Competitor Watch
By Shimecki.
This template keeps a list of public pages to watch and compares them with saved snapshots. Use it when you want to review a meaningful change to a named page.
Start with one public URL, your own product context, and the kind of change you care about. Ask it to save the first baseline and show where that snapshot will be kept. On a later run, check the old and new wording against the sources before changing your own offer.
The listing describes recurring routines as disabled at the start. Review the first results before choosing a schedule. An unreadable page or missing earlier snapshot stays a gap in the comparison.

02. Copy Humanizer
By Massimo De Luisa.
Got a draft that takes forever to get to the point? Copy Humanizer is designed to revise your copy while preserving its facts and voice, with an explanation of the edits.
Give it a paragraph from a campaign email, your intended reader, and the words or claims you want to keep. Ask for the revised paragraph and a short change list.
Edit this paragraph for clarity. Keep my facts, meaning and voice. Show the revised paragraph and explain the changes. Keep these words or claims unchanged: [list]. Flag anything that needs a source. Don’t add facts or promises.
Read the versions side by side. “Your order has shipped” should still mean shipped, with no invented arrival date. Keep a line that sounds like you when it helps the message.


1 Agentic Alpha
Interview First, Plan Second, Execute Third
One of the biggest “hacks” I’ve learned in my years of working with Super Intelligence™, it’s this:
Don’t try to Chad-ram it to do what you want without yapping (or typing, you filthy animal) for 5-10 minutes and get your ideas out of your head through an extensive interview (or grilling) process.
My favorite SKILL for this is made courtesy of Matt Pocock. Get that here. Take this link and drop it in your Chief of Staff Grok Bot and have it interview until there is shared understanding of what you want, why you want it, and what good vs bad looks like.
Agents are good enough now to figure out the “how” for most knowledge work.
After your interview, have it craft a plan.md file. There are various skills that can do this, but a simple markdown file that turns what you just discussed in the interview stage into an implementation plan is good enough.
Once you’ve got that Plan® file, tell your Grok Bot CoS to delegate it to the appropriate Grok Bot on your squad. If you don’t have the right teammate, go ahead and use dr. eggbot to build a custom Grok Bot that meets your needs.
In closing, there are more agents to try, and I'm going to keep trying them.
As the infamous William Montgomery of Kill Tony, the #1 Live Podcast in the World, says:
I AIN’T NEVA GONNA STOP.
Reply and tell me which agent you're using, one job you've handed it, and the part you still want to check yourself. I'm curious where it's actually helping you.
Toodles ✌️
Jesse
P.S. Remember, your taste, instincts, and wisdom still come through when you decide what to delegate to your Grok Bot squad. Never outsource your soul. That’s for the lames, and we don’t subscribe to that ish around here.
How was this issue?
| |||
Pass it along.Share this issue ↗Send to a friend ↗ | |||




