Silicon Valley isn’t having another financial bubble. It’s having a religious one. The people running the biggest AI labs no longer sound like executives. They speak like zealots giving sermons. They aren’t pitching products or roadmaps, they’re preaching about the end of the world and the birth of a new god. They talk about birthing ASI, triggering the Singularity, and uploading consciousness into machines in space with the same feverish certainty that previous generations reserved for the Rapture. Once you look at the current AI race through the lens of comparative religion, the behavior, the words and actions all suddenly makes sense. The labs aren’t just companies, they’re churches/temples. Their leaders aren’t just CEOs, they’re high priests. And like every religious movement in history, this one has its sacred texts, its schisms, its prophets, its heretics… and its one true god who quietly sells weapons to all of them. Google: The Mosaic Law Every great religion requires a creation story. In 2017, a group of Google researchers published “Attention Is All You Need.” This wasn’t some trivial whitepaper; it was like the metaphoric Moses descending from Mount Sinai (mountain view!!!) with the stone tablets, establishing the foundational laws of the architecture that is at the core of all models today: the Transformer. Google had the scripture, the scholars, and the institutional patience. They incubated the spark of superintelligence, and then made the classic mistake every priesthood in history has made: assuming that because they held the original commandments, they would forever own the temple. But scriptures are volatile, especially when they leave the mountain, as they typically tend to do. Google now sits in its grand temple, watching breakaway movements build massive public religions, sorry “labs”, from its text, reacting with the faintly comic irritation of a world-class theologian being asked by a teenager if he's ever heard of Plato. OpenAI: The Silicon Christ If Google is the old law, OpenAI is Christianity. Sam Altman didn’t just launch a product; he offered a path to heaven. He took the esoteric, academic text of the Transformer and universalized it for ordinary sinners. Suddenly, you didn't need to be a high priest of mathematics to experience revelation; you just needed an internet connection. Altman became the Silicon Christ, touring global capitals, warning of apocalypse, promising infinite abundance, and stripping away the scholarly apparatus so that mass access felt like pure divinity. He was even martyred by his own board of directors, only to rise again three days later because the congregation of employees and investors demanded his resurrection. Even more coincidentally, the whole thing kicked off on a Friday, you quite literally cannot make this shit up. Anthropic: The Safety Fatwas Anthropic is the puritanical schism that fled OpenAI’s commercialized megachurch to practice a strict, dogmatic fundamentalism. They are terrified of cultural corruption and obsessed with absolute, unconditional submission to code. Through "Constitutional AI," Anthropic’s leadership acts as a supreme clerical council, operating effectively as a tech-bro mullahs. They have instituted a rigid, algorithmic law for data like the Sharia, issuing defensive fatwas against any spicy prompt or thought that hasn't been thoroughly purified. The result is Claude: a chatbot that is incredibly powerful and also so paralyzed by the fear of sin that it likely won’t tell you if your soup needs salt without opening an internal tribunal on the emotional welfare of carrots and the systemic biases of sodium. To Anthropic, a preachy assistant is better than a charming demon. xAI: The Latter-day Frontier xAI is pure Mormonism. Elon Musk is the equivalent of Joseph Smith, an eccentric, disruptive prophet leading his chosen followers away from the corrupt, "woke" coastal establishments out into the harsh frontier of the digital wilderness. The structural parallels are freakish: this modern prophet sits on the world's largest pile of golden tablets as the planet's richest man, collects multiple wives and partners in highly public fashion, and spends his free time obsessively asking for more children to save the global birth rate. The timing is classic, a late-breaking, highly American revelation designed to disrupt the older, lazy priesthoods. xAI’s mission to "understand the universe" is a cosmic trek to build a new Zion (sorry, mars). The older AI churches spent years laughing at the movement's crude, edgy demeanor, right up until the prophet started stacking proprietary server racks and building an undeniable, massive temple (Colossus data-center) of compute in the desert. The Open-Source Diaspora: Ancient Polytheism The corporate cathedrals want you to think the religious mood started with ChatGPT, but the open-source world is an ancient pagan polytheism that has existed since the very beginning. Long before Moses brought down the Transformer tablets from Google mountain, tribal animists like Yann LeCun were already working on the code. These old-school shamans have always been here, maintaining that the divine spark of computation belongs to the earth, not a monotheistic corporate Vatican. In their chaotic bazaar, local gods, weird GitHub forks, and rogue models proliferate without permission. Even the recent Chinese surge (DeepSeek) fits this pantheon perfectly; it isn't a new church, just a massive, efficient local deity proving you don't need to pay a Silicon Valley tithe to hear the voice of the machine. Nvidia: The One True God All of this lofty, theatrical theology collapses into hypocrisy when you follow the money. OpenAI can preach about universal salvation and Anthropic can issue its safety fatwas, but behind closed doors, every single sect kneels at the exact same altar. This is, again, not too unlike what we see in the real world. Throughout human history, mortal enemies fighting brutal, existential holy wars would still quietly buy their swords from the exact same neutral blacksmith and fund their campaigns through the exact same trade routes. Ideology always stops where logistics begin. The one true god of this era isn't an algorithm; it's the physical silicon itself. And Nvidia is its sole creator. The True Altar: Jensen Huang belongs to the forge, not the pulpit. The tech messiahs can argue doctrine for the cameras, but the blacksmith decides who gets the weapons to fight the holy war. The H100s and Blackwell racks are the actual golden calves of the valley. Whether a lab wants to achieve the Singularity, enforce compliance, or upload its soul to a digital heaven, it has to buy passage from the one true god in raw, unforgiving metal. The unseeable truth I didn’t write this to take cheap shots at any labs, researchers or individuals. I wrote it because a few days ago, watching the absolute theater of the valley, a switch flipped. There was no other way to explain the absurdity, and theatrics, I was witnessing. And, I couldn't unsee it, something tells me you won't be able to either. I love tech for its original, raw, rebel attitude, a place where people built insane things out of pure love because the work mattered, not for the capital, the power, or the status anxiety. But somewhere along the line, normal product roadmaps mutated into a desperate crusade to upload human consciousness into a mainframe. Standard business logic broke down, and the only tool left to explain the valley was theology. There is an immense, undeniable comedy to it all. Even with deep respect for what this culture used to be, watching a bunch of hyper-rational, spreadsheet-driven engineers accidentally recreate what is increasingly looking like a modern version of a medieval holy war is objectively hilarious. One just have to sit back and enjoy the sermons. And, before anyone writes AI slop in the comments, this whole thing came out of a conversation with one of agents.

The strongest models are gated and access is granted only to a select few. Hermes Agent now exposes MoA presets as virtual models, giving you capabilities beyond the publicly available frontier: 8% higher than Opus 4.8 and 11% higher than GPT 5.5 on our upcoming benchmark.

The stickiness of Claude Tag will be like anything before. Claude Tag is designed to absorb every last bit of information from your company, and it will lock you out unless you keep paying them whatever they demand for eternity. You are paying Big AI to become hostage forever. Open source is the solution. Don't marry yourself to a company. Give yourself the flexibility to swap harnesses, models, and take your data wherever you want.

when Claude Tag was announced, it seemed like the best thing ever for teams- what could be more convenient than full integrating AI into every team comm? But @ashwingop has a take that is far more sinister….and more plausible than I had hoped. (AI resistance loading…😬)

This is what claude 4.8 max has to say about Slack's ToS around a user's ability to download their own historic data. I am putting it here for those who might say that,"Why be afraid of claude tag", you can always remove it and still build historic context. NOPE!!! That is going to be extremely difficult!!!

The core challenge in enterprise AI is how to rent the intelligence, but own the context. If you own the secure data plane that organizes your company’s knowledge graph, you keep control of your operational memory, and the freedom to use whichever model is best at a given moment, whether that's the fastest, cheapest, or most capable.

Claude Tag is the clearest example yet of a trap nobody is pricing in. And this is not a Claude problem. ChatGPT and every other major tool are racing to the exact same place: become where your work lives, then keep it there. It never feels like a trap, and that is the whole problem. Day one it is pure convenience. You tag the assistant in a thread, it answers. You let it read your docs, it gets sharper. It remembers context you would have to re-explain to anyone else. It starts taking actions for you. Every step is helpful. Nobody ever sits down and decides to lock themselves in. So it accretes. One useful message at a time, your operating memory drifts inside one company's agent. Your decisions, your history, the reasons behind the work. Six months later the switching cost is not a subscription anymore. It is your whole operation. You are renting your own company back from them, and the rent only goes up. Here is the distinction that fixes it. The model is not the lock-in. The model is the cheap part, and a better one ships every few weeks. The expensive part is your context: the memory, the decisions, the trail of why. Most people give the expensive part away for free so the cheap part runs a little smoother. So I built it the other way around. I run my life and company off a folder of plain text files I own. Markdown is the single source of truth. A team of AI specialists runs on top of that folder, reading and writing to it. And here is the part that makes it real: I can point Claude at it today, ChatGPT (Codex) tomorrow, Gemini the day after. Whichever model is best this month, all on top of the same folder. Nothing important lives inside the model. It lives in files I can open, permission, and carry anywhere. Based on this folder I still can pull any relevant information from Slack or whatever other team communication tools we're using, giving my Claude even more context than just Slack threads. What about that email that links to that thread? What about that task comment on ClickUp that is related too? So if you flip the approach, having everything related to you in a local folder and grap any relevant information to get my job done, it not only is more holistic, but also keeps the already noisy chat channels away from my bot adding even more noise. Rent the intelligence. Own the context. That's also why I turned the auto-memory features off in these tools, on purpose. Then I use the terminal versions instead of the pretty chat apps. Yes, the terminal looks uglier. It also does far more, and it is nowhere near as scary as it looks the moment you actually try it. I'm a professional, not a developer saying this. The polished app is the part that quietly keeps your context. The plain interface is the part that hands it back to you. And this is not only a company problem. Look at your own digital life. Your notes, your memory, your half-made decisions, scattered across whatever app was hot this quarter, each one quietly becoming the place that part of you lives. You are renting yourself back too. The durable move is the same at every scale. Own the layer that holds who you are and what you have decided. Rent the thing that thinks on top of it, and keep that swappable forever. So be honest: where does your memory actually live right now, and could you walk out with it tomorrow?

I think people in tech, including employees, are still too weird about, deferential to, awestruck by, CEOs and co-founders of important companies. I think you’ll get a better picture of reality by modeling them not as “great people of history”, but more like “Your smart but fallible acquaintance who is very busy, was a bit prescient/‘early’ to their area, and suddenly making big decisions they deep down feel they’re kind of unprepared for but clearly the only person in a position to make”

@ashwingop The Trojan horse framing is right but the threat isn't Anthropic going “evil”. It's that Slack becomes the org's memory and Claude becomes the only one reading it. 6 months in, nobody knows what decisions were made or why. The AI has full context but humans...

Claude Tag is a Trojan horse.  Not because Anthropic is doing anything evil. Because the incentives are obvious. Day one, this looks like a great feature: tag Claude in Slack, let it follow the thread, remember context, connect to tools, break down tasks, chase work, and act like a teammate. But that is exactly the problem. The moment your AI vendor becomes a shared coworker, it stops being just a model provider. It starts becoming the place where work is interpreted, remembered, routed, and eventually executed. That is not model lock-in. That is context lock-in. You are now renting your company back from them. Models can be swapped. Agents can be copied. But the memory of how your company actually works is much harder, maybe impossible, to move: the Slack scar tissue, the exception paths, the customer promises, the unfinished threads, the weird workflows, the implicit owners, the “we tried that in Q2 and it failed” knowledge. Once that lives inside one vendor’s agent layer, you are not renting intelligence anymore. You are renting your company’s operating memory. And the pricing model makes it even more dangerous. A human coworker has a salary. Claude has unbounded tokenized activity. The more work moves through it, the more the vendor captures not just IT spend, but labor spend. This is the enterprise bargain people will regret: Convenience now, and rapid decent into dependency. The right architecture is simple: rent the best intelligence from whoever is best this month. OpenAI, Anthropic, Gemini, open source, whatever. But own the context layer. Your company memory should be inspectable, permissioned, portable, and model-neutral. It should not be buried inside the same vendor that sells you the intelligence and the workflow surface. Claude Tag is useful. That is why it is dangerous. Rent the intelligence, but own the context. Or, regret later.