AI gen & optimized visual sitemaps for websites. Run Audits ➡️ AI Optimize IA + Content 🧠 ➡️ Create Flows for SEO, UX, Marketing, and Competitor Analysis
New York, NY
Joined January 2018
- Tweets11.6K
- Following631
- Followers732
- Likes7.2K
#Vibe-coded the site ✅
Now make its structure work for SEO and AI search. Turn a crawl into a visual sitemap + AI audit report w/ optimizations and full report. Get it for a LIFETIME.
app.visualsitemaps.com/lifet… #saas #LTD #appsumo
#Vibe-coded the site? Now make its structure work for SEO and AI search.
Turn a crawl into a visual sitemap + AI audit report w/ optimizations and full report.
Get it for a LIFETIME.
app.visualsitemaps.com/lifet…
#saas #LTD #appsumo
VisualSitemaps retweeted
Good Will Hunting GPU Edition
Ep. 2: My Boy's Wicked Smaht
Ben Affleck reveals he writes Python, understands convolutional neural networks, worked extensively with GPUs, and used his celebrity status to get private looks at Google and OpenAI’s video models
“I’ve always been kind of into computers since I was young. Then, when film started to move from analog film to digital, I became more interested in that aspect of it. The visual-effects workflow for many years has included machine learning, so I can write pretty shitty Python scripts and stuff like that.
“With convolutional neural networks, which were the precursors to what the transformer can do, which is much more computation simultaneously, you would do things like look at what’s called a tensor. That’s the numerical translation of a visual image in numbers, like the batch number, the frame number and the red, green and blue values of each pixel in each frame. It’s just that simple. That numeric is called a tensor.
“You’d use a convolutional neural network to identify patterns that reveal what’s called edge detection or feature extraction, which is identifying patterns well enough to know, this is where the window ledge is, so we can more easily take the green-screen image out and replace it with something.
“That was familiar to me early on because, prior to Artists Equity, I had a small visual-effects company. I’ve worked with GPUs a lot too. The visual-effects guys said, ‘Hey, you should see. There are a couple: Google and this other company, OpenAI, are doing really interesting stuff with transformers in video.’
“I’ve learned that I can actually just call up and go, ‘Hey, it’s Ben Affleck. Can I come see what you’re doing?’ Sometimes people say yes, to my astonishment.”
1/ I just spent [~a week] getting an AI "cofounder" running on my VPS — #Hermes agent + a #gBrain knowledge layer.
The whole time, there was a guided installer that does it in ~30 min.
I did it by hand instead. Here's everything that went wrong so you don't.
2/
The setup: Hermes (a Docker-based agent runtime) on a Hostinger VPS, with gBrain on top — a Postgres-backed second brain that does hybrid search + a knowledge graph and consolidates memory overnight.
Goal: an agent that checks institutional memory before it answers.
3/
The mistake: gBrain ships a guided install. You paste a URL, the agent clones it, sets up the DB, loads the skills, wires the dream cycle + autopilot, and verifies the whole thing.
I skipped it and installed manually — piece by piece, troubleshooting as I went.
4/
Why that's a trap: every manual step makes sense in isolation. Install this. Configure that. Add a relay.
You never stop to ask "is there a paved road that does all of this at once?" Local optimization, global swamp. I ended up with a bare brain, zero skills, no automation.
5/
The biggest landmine: gBrain defaults to PGLite, which allows ONE process at a time.
So the MCP server, the autopilot, and the nightly dream cycle all fought over a single lock. Whoever held it blocked the others. No reinstall fixes this — it's a hard limit.
6/
The fix: move off PGLite to real Postgres + pgvector (self-hosted on the VPS, or hosted like Supabase). Postgres handles concurrent connections natively, so every process runs at once.
If you're doing anything beyond a toy brain, start on Postgres. Don't discover this later.
7/
VPS pain #1 — Docker networking.
gBrain ran on the host; the agent ran in a container. They couldn't see each other. I needed a socat relay to bridge the Docker bridge IP to localhost, plus a host-gateway entry on the container. Headless VPS makes every bit of this harder.
8/
VPS pain #2 — the silent config killers:
a poisoned credential pool that survived restarts (fix: nuke auth.json, set creds as a custom provider)
a wrong HOME inside the container that made MCP discovery fail with no error
one stray toolsets: line that hid every MCP tool
9/
VPS pain #3 — caching + retrieval gotchas:
the web UI caches which MCP tools exist at startup, so after any outage you must restart it to see tools again
the agent's get_page only reads the default source — pages synced elsewhere are invisible to it
after migrating DBs, you must re-embed
10/
Meta-lesson #1 (the expensive one): on a headless VPS, use the guided installer and start on hosted/self-hosted Postgres from day one.
Hand-wiring socat, systemd, credentials, and Docker networking is where the days go. The paved road exists specifically to skip that.
11/
Meta-lesson #2: ask if you even need it yet.
At ~15 markdown files, the agent reading the files directly beat hybrid search + a knowledge graph. That infra pays off at hundreds/thousands of pages, not fifteen. I built the warehouse before I had the inventory.
12/
Meta-lesson #3: when you're debugging with an AI step by step, periodically force the question "is there a one-shot install for this WHOLE thing?"
The model will happily help you optimize each manual step. It won't always stop you to point at the elevator next to the stairs.
13/
Where I landed: a working brain on Postgres, embedded and retrievable, with the dream cycle on a nightly timer and auto-restart watchers so it survives reboots.
Good outcome. Just took the scenic route. If you're wiring an agent + brain on a VPS — take the installer.
Mythbusting generative #AI #search: what you don't need to do - Google May 2026
As generative AI search evolves, so have the theories and practices—and sometimes, the misconceptions—surrounding it. While terms like Answer Engine Optimization (#AEO) or Generative Engine Optimization (#GEO) are common online, many suggested "hacks" aren't effective or supported by how Google Search actually works.
To help you focus on what matters for your website's visibility, we've collected some of the most prominent topics circulating the internet around generative AI and Google Search. Here are a few things you can ignore for Google Search:
LLMS.txt files and other "special" markup: You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in generative AI search. Note that Google may discover, crawl, and index many kinds of files in addition to HTML on a website: this doesn't mean that the file is treated in a special way.
"Chunking" content: There's no requirement to break your content into tiny pieces for AI to better understand it. Google systems are able to understand the nuance of multiple topics on a page and show the relevant piece to users. However, sometimes shorter (or longer!) pages can work well depending on your audience and subject matter. There's no ideal page length, and in the end, make pages for your audience, not just for generative AI search.
Rewriting content just for AI systems: You don't need to write in a specific way just for generative AI search. AI systems can understand synonyms and general meanings of what someone is seeking, in order to connect them with content that might not use the same precise words. This means you don't have to worry that you don't have enough "long-tail" keywords or haven't captured every variation of how someone might seek content like yours.
Seeking inauthentic "mentions": Just like the rest of Google Search, our generative AI features can show what's being said about products and services across the web, including in blogs, videos, and forum discussions. However, seeking inauthentic "mentions" across the web isn't as helpful as it might seem. Our core ranking systems focus on high-quality content while other systems block spam; our generative AI features depend on both.
Overfocusing on structured data: Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add. However, it's a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search.
developers.google.com/search…
Most sitemap tools show what a site looks like today.
We built AI that shows what it should become.
VisualSitemaps AI generates:
✅ optimized future-state sitemap
✅ topical SEO clusters
✅ IA/SEO strategy report
For UX, SEO, content, and web teams 🚀visualsitemaps.com/ai-optimi…
VisualSitemaps retweeted
Jensen Huang, NVIDIA CEO: "It even runs large language models" — a $249 AI computer that fits on your desk.
"Picking Brains" will soon be ubiquitously productized.
"Thought Leaders" & Influencers will soon be dumping their whole corpus into a rented GPU and offer TimeShares 🤓
