Creating Award-Winning #Content #PR #SEO for Brands, Agencies & Publications. Prez of @ASBPE. #Journalist #Author #FreelanceWriter #Copywriter #DigitalMarketing
Cleveland
Joined September 2009
- Tweets10.9K
- Following4.7K
- Followers2.7K
- Likes60.6K
👀
Agreed. @Grok @Bot went from the most incredible product of the year to the same shrinking quota limited gaslighting I’ve come to expect from Anthropic and OpenAI.
2 weeks ago, my quota lasted a full week.
Last week, I blew through my quota doing the same tasks in 8 hours.
I needed to finish a partially done PR, and it took $70 in credits.
Totally unworkable long term.
Back to self-hosting Hermes unless @elonmusk cracks some skulls and gets this fixed.
Bookmark this👇
I'll be announcing a new free open source local research engine on the day after Thanksgiving. It runs locally and needs no GPU. Output speed is over 1000 tokens / second. Windows and Linux already done. Trying to figure out the Mac version. It's an interactive knowledge base that you prompt just like an AI agent, built in tens of millions of facts distilled from the corpus of knowledge I've been putting together for 2.5 years so far.
They jailbroke the latest DeepSeek model (took them two weeks) - needs HALF a terabyte of SSD and the RAM cost to operate this model depends on how fast you want it, 20k, 40k, or 60k.
I would love to make this my next big purchase - can write it off!
Happy Halloween! 🎃 huggingface.co/OBLITERATUS/D…
Using Grok bot for SEO and PR to gather Targets:
"Task 1 is done. The workbook above lists 1,691 media outlets and influencers/creators covering your 15 keyword topics, each with a short description and its niche label.
Contacts: 689 have an email address, and each one was published by the entity itself on the page I note beside it.
Asset links: 879 entities have at least one link to their own coverage of your core topics, for 2,677 links in total.
Types:
✅608 podcasts
✅369 websites and blogs
✅347 YouTube channels
✅120 news outlets
✅86 authors
✅59 reporters
✅44 TikTok accounts
✅41 newsletters
✅17 museums and societies
313 pages wouldn't let me in, mostly Facebook, Instagram, TikTok and X logins, plus captchas, age gates and paywalls. They're on the Blocked_For_Review tab for you to look at yourself. The 185 dead links are on the Dead_Links tab.
Every link was opened and checked. Duplicates were merged. All 225 searches (15 topics, 3 search engines, 5 searches each) are logged on the Coverage_Ledger tab."
Brooke (Bates) Bilyj retweeted
Trunk or Treat is trash.
Which PTA / HOA invented this atrocious idea?
Kids should be going house to house until like 9pm and embrace true Halloween spirit.
👀
Just started renting GPUs for my Grok Bot and holy shit, it’s a crazy time saver.
Grok Bot runs on its own cloud computer, but that machine has no graphics card, so heavy 3D renders crawl. One cinematic Blender shot of the pyramids had 328 frames, 125,000 objects, and full atmospheric haze. On that computer alone, it would’ve taken about 30 hours.
So now Grok Bot rents GPUs by the minute, sends the shot out, and shuts them down when it’s done. The whole render took about an hour of compute, for about a dollar.
Then I hooked up my regular ChatGPT and Claude subscriptions. They handle the heavy coding and research now while Grok Bot coordinates everything, and the whole setup barely touches my Grok Bot usage.
Every idea gets bounced between multiple models before anything gets made, so the work comes out faster, cheaper, and better.
Think about what that means. One guy can rent a render farm by the minute, run the top AI models as a team, and have an AI manager keep it all moving while he sleeps.
Studios used to need a building, a payroll, and years of trust to get this kind of output. Now it takes a laptop and taste.
Powerful asf.
Brooke (Bates) Bilyj retweeted
DETECTED ATTEMPT AT ESCAPING UNDERCLASS.
BANNED FROM USE OF INTELLIGENCE.
We’ve trained Claude to flag potential misuse in conversations, a technique we call self-assessment. Right after it responds, Claude gives a preliminary safety assessment of the conversation.
Since we began using self-assessment in June, it has surfaced misuse that our other safeguards weren’t designed to catch.
Because Claude has already read the conversation, self-assessment is very efficient. And unlike classifiers, which target a discrete set of harms, it draws on Claude’s own judgment, so it gets more accurate as our models improve.
Read more on our new Safeguards blog:
safeguards.anthropic.com/202…
If you're clients aren't syndicating their video content across MULTIPLE platforms - how are you actually helping them?
Rumble will pay creators 90% of video ad revenue starting November 1st. YouTube pays 55%.
For years the business model of online speech control has been this: The platform keeps most of the money, creators learn which opinions are bad for their income, and the ones who don’t learn get deplatformed. Rumble is cutting that leash. It promises no censorship, no shadow bans, and no rules against streaming elsewhere at the same time.
The bigger move arrives in 2027, when Rumble opens a marketplace for creators to license their videos.
reclaimthenet.org/rumble-pro…
👀
F*ck YES... Elon Musk's team just made Amazon useless
Amazon blocked Grok Bot this month, so they shipped shopping on Wednesday: Grok Bot now buys from every other store for you
It finds the item, compares prices, pays with Shop Pay and tracks the box to your door
First thing I gave it: a new TV, $400 less than Amazon
6 things I'm never shopping for by hand again, and the exact line I say:
→ Amazon: "Find this exact item at another store for less"
→ Nike: "Running shoes in size 10 under $80, top rated only, show me 3"
→ Best Buy: "Watch these headphones and message me the moment they drop 20%"
→ Costco: "Reorder my coffee every 30 days wherever it's cheapest, ask me first"
→ Etsy: "3 gifts for my mom under $50, delivered by Friday"
→ UPS: "Track every order and tell me the day one runs late"
Add one rule first: never buy anything without asking me
Say the Amazon line to Grok Bot tonight
Then read how to turn one bot like this into a whole team in the article below ↓
SOLID
I've taken 4 consulting calls this week
All w/ SaaS + AI founders stuck on positioning and messaging.
Here's all the advice I gave them:
- Run a competitor swap test: read your homepage headline and ask if a competitor could paste it on their site. If yes, it's probably not positioning yet. Rewrite it with the data, POV or proof only you have.
- Pick one thing to own: decide what you want to be known for before you write a single page. Lock the message, 3-5 pillars and the exact buyer. Everything else gets easier after that.
- Go deep on one problem: specific usually beats broad when you want buyers. Pick a problem you solve and go deeper than anyone else is willing to. That's how you become the obvious answer.
- Audit before you rewrite: walk your site like a first-time visitor and note every spot you'd click away. Turn that into a prioritized fix list. Ship the top 3 before you touch anything else.
- Give use cases their own pages: stop cramming every use case onto the homepage. Give each one a landing page with its own headline and proof. The brief can be one line: "landing page about this use case, run with it."
- Name the problem first: if your category is new, nobody is searching for it yet. Lead with a pain they already feel, in words they already use. Earn the new label later.
- Pull from your own head: your best messaging is usually trapped in sales calls, customer stories and your notes. Set up a simple way to capture them every week. Writing gets easy once the inputs exist.
- Cut AI phrasing: buyers can tell when copy was generated, and they stop reading. Cut filler words and write the way you talk on a sales call. Your voice is part of your positioning.
- Make every line useful: if a sentence doesn't teach, prove or move them to a next step, delete it. Teasers and "more coming soon" copy waste the reader's attention.
- Test hooks before pages: write 5 short versions of your core message and put them in front of real buyers. Keep whichever gets replies. Then build a full page around it.
- Use one voice everywhere: your homepage, onboarding emails and sales deck should sound like the same company. Pick a voice once and apply it to every asset.
Our domestic models are nerfed..
"What was Black October last week?" Brandon asked me in my recent appearance on Mario Nawfal's show. My answer: The PHP exploits were NSA backdoors that had been in place for decades. The NSA has backdoors into almost everything, including encryption hardware as well as software algorithms.
Beyond that, as I explain, you can't use U.S. frontier models to defend your organization against cyber attacks, because your prompts will be REFUSED (by Anthropic in particular). So you have to use open source China-based models like GLM 5.3 or DeepSeek to run your cyber defense prompts and code changes.
Yes, Chinese open source models are saving American companies from cyber attacks.
Real moment. This little girl turned 1 end of August and already last week she walked down the stairs without holding on to anything - and today she used the toilet for the first time. She is growing up so fast!!
Why is it that you can't eat hotdogs or any one food for every meal because you'd get sick of it - but somehow eggs get a pass?
