Tag the website link on your Google Business Profile with a UTM. I use utm_campaign=gbp.
Search Console treats it as its own URL, so you can finally see Maps traffic separate from regular organic. Otherwise those Maps clicks blend into your organic numbers and the average position gets muddy.
Make sure your agent knows too, so it doesn't report those positions as organic rankings.
14
Astro builds static files, so I can host client sites on Cloudflare for free pretty much forever.
Same reason leaving is easy. If a client ever wants to go, I export the static site and hand it over.
18
Efficiency improvements per-task and higher intelligence for the same price are always welcome.
We seem to be past the time of rising model costs, for now.
Some of you have been asking very politely. Some of you have just been asking.
GPT-6.2 Astra and Sol are now available.
Super efficient and built to perform where it counts—the real world.
openai.com/index/introducing…
20
The most useful part of asking ChatGPT "best [service] in [town]" isn't which businesses it names. It's which sites it cites.
Those are the directories and lists it trusts for that area. That's basically your listings to-do list.
25
AI agents doing competitor research will tell you a competitor has no blog or no reviews when they really just couldn't find them.
I make mine write "not found after 3 checks" instead, unless the competitor's own site proves it, like a sitemap with no blog in it.
Unknown and zero are not the same thing and agents mix them up constantly.
16
Ask a local business owner who their competitors are and you'll usually get two lists mixed together.
Who they actually lose jobs to, and who they wish they were.
Only the first list belongs in your SEO gap analysis. Then add whoever keeps showing up in the Maps pack that they never mentioned. Comparing against the dream list just gets you a plan for a business that isn't theirs yet.
12
Patchstack counted 11,334 new WordPress vulnerabilities in 2025. 91% were in plugins.
That's a big reason my client sites are on Astro now. A static site has no plugins, no database and no wp-admin for bots to hammer. Nothing to patch every week.
3
1
22
If you blocked GPTBot to stay out of AI training, double check you didn't block OAI-SearchBot too.
GPTBot is training. OAI-SearchBot is what lets you show up in ChatGPT search. Claude has the same kind of split with ClaudeBot and Claude-SearchBot.
Easy one to miss if you blocked every AI bot a while back.
30
I've started writing agent instructions with the why before the how.
"Don't run builds on my Mac" only covers builds. "My Mac keeps crashing when agents run heavy stuff on it" and they figure out installs, test suites and everything else heavy goes to my VMs too.
The reason ends up covering a bunch of stuff I never would've thought to write a rule for.
1
37
I moved my client sites from WordPress to Astro, and speed was honestly the smallest reason.
The big one is that every client site is built off one template. The structure and SEO stuff live in the template, and each client just brings their own content and styling.
So when I make an SEO improvement, every client gets it. No logging into each site to update them one by one.
3
1
41
Try asking your agent in T3 Code to "delegate this to my vps."
It shows a card where you pick the machine and how much access it gets, then the task runs there and reports back. Linked environments is still a beta, so turn it on under Settings > Labs on both machines first.
1
1
28
Long term I think Claude will write the instructions for Codex and then review what Codex builds.
Right now they mostly check each other's work in T3 Code, and that seems to catch more than either one alone.
Has anyone settled on which model plans and which one builds?
4
3
56
What happens when a business fails verification on ChatGPT Business?
My business verification got rejected by Persona despite me submitting ID proof, with no rejection reason or path to reapply.
T3 Code has load balancing across machines and I'm about to turn it on.
Auto balance picks a machine for each new thread based on free CPU and memory. You can set each one to Prefer, Less often, or Manual only. With a few VMs running, that beats picking one every time.
8
Claude Fable 5.2 coming out 20% cheaper than 5.1 is awesome, but I feel like I’ll still burn through it in minutes on the $100 plan.
What are the best uses for Fable that really make it worth using over Opus?
1
37
Next thing I'm setting up in T3 Code is linked environments.
My VMs aren't signed in to my deployments or MCPs. The plan is for my agents on the VMs to hand anything that needs that access to my main machine, while most of the building stays on the VMs.
1
2
35
Try T3 Code if your Claude and Codex subs live in separate apps.
Just ask Claude Code or Codex to set it up for you. Then switch between them from the model picker, or have them call and delegate work to each other.
1
40
There’s a reason why this man’s products are among the most popular on earth.
Just did a round table review with the Grok @Bot engineering team.
Fixes are being implemented overnight and you should notice a dramatic improvement by tomorrow. If not, reply to this post and we will address any remaining issues right away.
31






