This is 100% the move.
A centralized, up to date knowledge base that all your agents can access so you're never locked in to one platform.
Here's how we built ours:
1. The whole thing is 2 private Github repos of plain markdown files. One company brain, one personal.
2. We numbered each folder by function. Company has brand, customers, products, sales, ops, finance. Personal has health, investments, family.
3. One AGENTS. md file that every agent reads first.
4. At the end of each day our agents run a "knowledge base audit" skill that compares the day's work to both brains and suggests edits.
What I love about this system is that it allows us to test all the new agent platforms without starting from scratch.
Zero agent lock-in.
When Grok Bot gives you exactly what you wanted, don’t stop at “perfect, thanks.”
Turn that result into a standard for the next run.
Try out this prompt I've been using: 👇🏻
“This is a strong result because [explain what made it useful].
1 - Extract the specific requirements that made it work.
2 - Separate reusable requirements from details that only apply to this task.
3 - Propose an update to the relevant workflow instructions and show me where it belongs.
4 - Keep this output as a reference example. Future work should use fresh evidence.”
For example, a useful morning briefing might give you:
The three decisions you need to make.
The deadline for each.
A recommendation with supporting evidence.
If that’s what makes the briefing valuable, capture those requirements.
“Great briefing” leaves the standard vague.
“Lead with decisions, deadlines and evidence” gives the next run something concrete to follow.
Your best outputs are examples of what good looks like in your business.
Use them.
One of the freelancers I'm coaching asked me: "What's the path from freelancing to really 'running a business' look like? (I ask because I'm already seeing that the time it takes to deliver on client work will be a future bottleneck.)"
Here's what I said:
Most freelancers take four paths:
1. Agency. This is what I did, obviously. You systematize your process and hire other people to implement it for clients. Great cash flow, can have enterprise value if you build it right. Can be a pain in the ass in terms of managing people AND clients... but these are problems that can be solved, you just have to WANT to solve them.
2. Selling courses/info: I did this. It's cool. Really solid leverage and cash flow. There's a right and wrong way to do it. The way I did it was the WRONG way. I became the linchpin for everything and I was selling "me" instead of a "methodology." @mrollyrichards did it correctly. He built Storylearningaround his system for learning a language... instead of him being the star of the show. Info can be good... but ive always felt its simply a matter of continuously having to launch "new shiny objects" which can get exhausting after a few years.
3. Launch your own brand: for example, it could be a DTC brand, where you use your skills to build a brand you own equity in. you are your only "client", at least while you're getting started. If you've read The E-Myth, you know that opening a bakery isn't about baking pies. In fact, once you start hiring, you stop baking pies and have to focus on running/growing the business. So it is a completely different skillset. You'll probably always be invovled in the marketing and sales creation on some level. But isnt your main job anymore.
4. Become an intrapreneur/partner with someone: partnerships can be ugly. they can also by dynamite. Kind of a roll of the dice. You can also join a company that's growing and work inside of it, possibly get equity and/or some kind of performance comp/stock/perks. It's not "true" Entrepreneurship... but I bet if you asked some of the early employees at Nvidia or SpaceX how they felt, I think they'd feel pretty happy with their decision.
Perhaps there are more, but these are the typical paths I see.
(pic somewhat related, but not really, just wanted to get you to stop scrolling. Thank you for your attention to this matter!)
Connect your ChatGPT and Claude subscriptions to Grok @bot with the prompt below (@elonmusk approved):
Connect my Claude and ChatGPT subscriptions so all of my bots can use them.
1. On your computer, install the Claude Code CLI and the OpenAI Codex CLI (upgrade Node if needed). Start each subscription login (Claude: `claude setup-token`; Codex: `codex login --device-auth`) and send me the sign-in links and codes. I'll sign in and paste back anything you need. Store the Claude token privately and load it automatically.
2. Set the defaults to these models: Claude Opus 5.5 for Claude Code and 6.1 Sol for Codex. Test each one with a quick prompt and tell me which models answered.
3. Add these MODEL ROUTING RULES to every one of my bots' system prompts, including yours. Keep everything else in each prompt exactly as it is:
- Writing (drafts, emails, posts, replies, copy, briefs): write the final text with Claude through the CLI, then review it against the bot's voice rules.
- Coding, scripts, data processing, and terminal work: use Codex.
- Browser and desktop tasks: keep using your normal computer tools.
- If either CLI fails, do the task yourself and say so.
4. Save the same rule to my shared memory so any bot I create later follows it too.
5. When you're done, tell me which bots you updated and confirm both tests passed.
.@klaviyo Composer is an "AI marketing agent that helps you build better customer relationships."
But what does it all mean!?
I'll tell you:
Every brand and agency on Klaviyo now has an AI chat interface built right into the platform that can:
- Answer plain-English questions about your data (“Where’s the biggest drop-off between first and second purchase?”) with citations you can verify.
- Benchmark your results against brands of similar vertical and size, drawn from 205,000+ Klaviyo customers. No more guessing whether a 2% click rate is good.
- Diagnose why engagement dipped last week, which flows are actually driving revenue, and where your segments and forms are leaving money on the table.
- Run recurring tasks, like a review of your top flows that flags anything slipping. These run in the cloud whether or not you’re logged in.
- Save your team’s process as “Skills” so it runs the same way across brands, markets, and team members.
- Audit live flows for expired coupons, broken links, out-of-stock products, and underperforming emails, then hand you a ranked fix list.
- Mine purchase history for product pairings and repurchase timing your segments don’t cover, then build the new segments you approve.
Here’s my take as someone who’s spent the last decade operating inside of Klaviyo accounts:
Composer won’t replace strategy, creative judgment, or knowing a brand’s core customer. But it can be an analyst that never stops working, surfacing opportunities that would be nearly impossible to catch through manual review.
If you’re on Klaviyo, open Composer today and ask it to audit your live flows. Will you love what it finds? 👀
What’s the first question you’re asking it?
HOW TO CONNECT YOUR GROK @bot TO WHATSAPP
send this tweet to your Grok Bot so it can handle the setup for you:
1. have it open web.whatsapp. com on its computer.
2. have it choose “link with phone number” and enter the number it already has for you.
3. have it generate the linking code and show it to you.
4. on your phone, open WhatsApp → Settings → Linked Devices → Link a Device → Link with phone number instead. enter the code to approve the connection.
5. have it check that your chats load and confirm it's connected. your Grok Bot can now read messages and draft or send replies on WhatsApp.
once connected, you can have it:
> organize your chats
> flag messages you haven't replied to
> add plans to your connected calendar
> and remind you about things you said you'd do
so between WhatsApp, my email, and X DMs, my Grok Bot now has an absurd amount of actionable context on my life :)
You can have Claude Code build a competitor ad library tracker and spy tool in minute snow for for Facebook ads, Google ads and LinkedIn ads.
You just need access to Apify for scraping.
And store all of your ads in an S3 bucket.
Then have Gemini 2.5 flash mini analyze the ads to understand everything that's going on and write this information to a postgres database.
Give your team access to this Postgres database via an API so that they with their claude code can query against this.
And then they can use this to inspire them for new ad creative that they should make.
Have this run weekly on a cron job so that netnew ads get loaded into the ad spy tool.
And then make this new ad creative using the chat gbt image 2.5 and seed dance endpoints.
And you can build this whole micro app and workflow on Graphed
pay as you go, only usage based
get started for free - https://t.co/HYtVDRggr8
[Grokbot update]
The best tip I learned about using agents is this:
Simply tell it your end goal and let it figure it out -- don't try to prompt it yourself
Example, I wanted an end-to-end system for our email copy
So, I told Grokbot that's what I wanted + described our processes + connected all needed apps
Now, once an AM approves email copy in the Grokbot chat, it:
- adds to google doc with proper formatting
- updates my table of contents at the top
- grabs the link from the TOC
- builds the asana task using our email template
- adds the proper send date to the asana tasks so my
automations fire with due dates for the team
- moves the task to the proper stage (design/client approval/etc.)
This will cut down our time ~50% for every single email we write, allowing us to focus more on strategy and high-level execution vs. task management + admin work
If you run agents, give each one a decision log.
It's one markdown file, six lines per run, newest entry at the top:
- Dated: the date and a one-line title for the job.
- Pulled: what the agent read, and where from.
- Decided: the call it made, and why.
- Acted: what it made or sent.
- Approved: whether you signed off, and what you changed.
- Outcome: how it turned out.
Then add one line to the system prompt so the agent keeps the file up to date for you:
After every run, add an entry at the top of decision-log.md in that order, include the reason even when it seems obvious, and leave Approved and Outcome blank until you know.
The line that matters most is the why in Decided. Six months in, you'll have a few hundred of them, like why you raised a price or why a proposal led with a guarantee, and you can paste them straight into the next agent you build.
Humanizer is the #2 trending model on Hugging Face and looks great and runs locally!
A 12B model that rewrites AI-written drafts (emails, essays, reports, forum posts; English and Chinese) so they read like a person wrote them.
Link: https://t.co/akc5rrOvdO
I just talked to a guy who makes $16,000 a month renting out a space he doesn't own.
He works a full-time tech job in Georgia, and his business has been open less than a year.
He paid $2,200 to rent a small venue for his wife's baby shower and thought, we could do this better.
So he signed a lease and cleaned up the space himself on nights and weekends with YouTube tutorials.
His best month was $22,500.
He averages about $16,000 a month, with about $8,000 of that profit.
No employees, just him and his wife.
He spent less than $100 on Facebook ads and booked about $10,000 in parties in one day.
Three churches pay him $1,000 a month each to use the space every Sunday. That covers his rent before he books any other party.
He also put QR codes on the walls. Guests scan them, upload selfies, and the photos pop up on a giant LED wall during the party.
Then he markets to everyone who uploaded, and 8% of them have booked their own event.
The event business is live and well here in GA.
In this episode Justin:
- Breaks down a $1,000 bare bones rental versus a $4,000 full package
- Shows the birthday mailers he sends 3 months before milestone birthdays
- Gives me his Facebook Marketplace trick for $30 tables
- Tells me why parking matters more than the building
This one was really good.
My Grok Bot limits ran out in two days, and I got tired of it
So I found a way out of this situation and compiled tips in a single article https://t.co/UebmdBgVSH
530 people asked for the system prompt I use to run SEO + AEO inside Claude Code and Cursor.
Instead of gatekeeping it in DMs, I just open-sourced the entire repo:
- Audits your codebase for technical SEO & broken metadata
- Reverse-engineers top 10 SERP titles & structure
- Writes comparison pages with real information gain
- Checks what ChatGPT and Perplexity cite for your niche
No $2,000/mo agency required.
Repo: https://t.co/7ZOsGjdJY0