Matt Cook

You can't trust doctors to know the full story today. Too much medical school propaganda. My doctor friends hate doctors, lol. Today you need other approaches based upon evidence.

I spent my younger career in direct sales. Then 15 years ago, started marketing. I had to learn everything. But the sales experience has let me become a good marketer quickly. Now I'm teaching AI to market. The "salesmanship in print" greats could market. Claude Hopkins ran an agency and built his own brand. Eugene Schwartz ran a number of successful offers in the market.

The gap between the advertisers who spent their careers mastering the ad channel button pushing and technicals vs. those who spent it mastering actual marketing principles is getting wider by the day.

Huge shoutout to @DrDaniS on a recent interview on @ChrisWillx YouTube show. Explains the low birth rate possibly. youtu.be/lsy41uI1dmY?si…

@EXM7777 I find clearing context and restarting is very helpful. You can't go over say 300K tokens without losing the thread.

Opus in Claude Code with 1M context is amazing... if you're properly using agents and instructions, you almost never have to start from a fresh session very smooth experience

@STR8HUSTLING @Tekeee Everything i buy is expensive now.

@CookResearcher @Tekeee What's the first thing you're buying,?

Gold is crashing. Silver is crashing. Crypto is crashing. Stocks are crashing. The dollar is crashing. Real talk what should we buy now?

@EricLDaugh When is this going to end? People I know who have traveled have reported very long waits.

🚨 BREAKING: Chuck Schumer has now unleashed SHOCKING chaos on airports nationwide A ridiculous amount of TSA agents are CALLING OFF the job Call-out rates: - 39% in New Orleans - 37% in Atlanta - 35% in Texas - 30% in New York - 10% nationwide NUKE THE FILUBUSTER, GOP! Go around traitor Schumer, and you become heroes.

5.4 is much smarter. It is the smartest model I’ve ever used. It is not afraid to contradict me and argue, either, and usually has been right. and Codex with 5.4 is faster and better than Claude Code Opus 4.6, at least for coding. I use Claude Code for other stuff, and love it, but it isn’t as good as Codex with 5.4 for writing code.

gpt-5.4 is a big jump over gpt-5.2 not in the fake "it feels smoother" way, but in actual reply quality. gpt-5.3 repeats my own words back to me and often still misses the main point i was trying to make but 5.4 is noticeably better at understanding what i actually mean

understandingai.org/cp/191398602 Will AI take many jobs away? This guy shows how we’ve been through this before, and we aren’t going to lose many jobs over AI. Is he right? I suspect he is.

It’s beneficial running your own agent in many cases. I am using Hermes Agent + OpenAI on a monthly plan for US$30/month. It’s nice to have our own agent interact with our meetings and run tasks. Claude Code is killer, and so Is Codex, but it’s great having an agent you can direct all the time and talk to on Telegram.

Claude is replacing OpenClaw and i’m honestly here for it

I’ve been using Hermes Agent with my OpenAI business subscription, US$30/month and so far it seems brilliant. With a good soul.md file it is brilliant and makes good jokes and has a great personality. I haven’t checked token usage but suspect it’s a fraction of OpenClaw’s.

@jonbrooks We are the early stages of a real estate bust. It’s hit the rest of Florida hard. When will it come to Miami? Overbuilding signs are everywhere in Miami.

Just got back from Miami… the real estate there is completely overpriced The numbers make no sense

Codex when it compacts doesn’t lose anything noticeable. When Claude Code compacts, you lose a lot. I store project state files on a regular basis during a session, so I can clear context and resume. With Claude Code, I have to do that. But with Codex, it isn’t really necessary.

Opus 4.6 v Codex 5.4. Codex is noticeably faster and smarter. I still use Claude Code where I want lots of interactivity. But frankly, Codex is the better harness and model and can do lots more than Claude in a smart way, without lots of intervention.

@TheCesarCross @glitch_ @NousResearch Can you be specific? What’s an example?

@glitch_ @NousResearch Private access could surface real edge cases under load

hermes agent by @NousResearch + my swarm + qmd = wild performance hermes on qwen with shared qmd outperforming stand alone opus 4.6 for me...

@CHRISM4FIA @Teknium @ErickSky @NousResearch There is nothing special about OpenClaw because it’s fairly easy to create similar, but better, agents. I think they bought OpenClaw to get Peter on board. Peter is a genius.

@CookResearcher @Teknium @ErickSky @NousResearch I still think OpenClaw has insane potential - even more with OpenAI behind it But it’s built ‘for everyone’ which has become its strength and weakness for me- some friends of mine have it deployed in their business already but for me- Hermes hits my mark

🚨 HERMES WORKSPACE YA ESTÁ AQUÍ El tocayo construyó el UI nativo que Hermes Agent (@NousResearch @Teknium) siempre merecía… ¡y con acceso móvil desde el celu! 📲 Esto trae todo: 💬 Chat con tools ejecutándose en vivo 🖥 Terminal integrada 📦 Browser de +90 skills 🧠 Memory viewer 📂 File explorer 🎨 8 temas (dark + light) ⚙️ Config completa de modelos Todo limpio, rápido y hecho para builders serios. Open source total, creado para el hackathon de Hermes. @outsourc_e se mandó una bestialidad con la que deberías aprovechar para usar este agente. ¡NO QUIERO QUE TE QUEDES ATRÁS! REPOOO👇

@CHRISM4FIA @Teknium @ErickSky @NousResearch No kidding. Openclaw was a maintenance nightmare. Hermes Agent just runs. Really enjoying it. A local model is powering one of my instances and it works pretty well too.

@ErickSky @NousResearch @Teknium Only had Hermes for ~24 now and as a small-large-scale one person business - it’s already had me feeling the weight off my shoulders of openclaw repair Love openclaw cool project- but reliability doesn’t hit the mark I need for business like Hermes does ⚕️❤️‍🔥🤖

One of my favorite Claude Code skills lets me put any written essay or letter and it looks at every single sentence, every single claim, and gives me citations for everything that I'm saying. It actually goes out and checks each citation so none of them are ever hallucinations.

@nickrgrs I get pretty good service in Florida actually. People are super nice here

the best service provided is always by people from the northeast it runs in their blood always on time, proactive, ahead of the curve, just excellent at what they do, etc etc florida, west coast, even parts of the south... service is totally shit in comparison

Agree or disagree? If it wasn't for Video porn, we wouldn't have a dating problem today

@PaulSolt We're so blessed I use both. I actually think Codex with 5.4 high is incredible and probably much better than Claude Code Opus 4.6 but I do use both. I like both.

GPT-5.4 vs. Opus 4.6 Which is writing most of your code?