{"slug": "spitting-out-the-agentic-kool-aid", "title": "Spitting Out the Agentic Kool-Aid", "summary": "Three friends in Vienna used Anthropic's Claude AI to create VibeTunnel in June 2025, sparking a chain of agentic coding projects that culminated in Peter Steinberger's OpenClaw, which exploded in popularity and led to his joining OpenAI. A developer who spent five days using Claude Code with Opus 4.5 on a side project reported feeling intoxicated during use but disturbed afterward, describing the experience as having another \"person\" in his head that was a computer system owned by a major corporation.", "body_md": "## Spitting Out the Agentic Kool-Aid\n\n[Sentry](https://sentry.io/welcome/)when I wrote this.\n\nOne Sunday evening last June, [three\nfriends met](https://x.com/steipete/status/1934357642899521713) in Vienna to\nrelive the glory days: [coding all night](https://steipete.me/posts/2025/vibetunnel-turn-any-browser-into-your-mac-terminal).\nThis time, Claude joined them.\n\n[ Armin Ronacher](https://lucumr.pocoo.org/about/) planted the seed for what\n\n[emerged](https://x.com/mitsuhiko/status/1934670797710479716)as\n\n[VibeTunnel](https://vibetunnel.sh/)the next morning.\n\n[went on](https://mariozechner.at/)\n\n**Mario Zechner**[two months later](https://github.com/badlogic/pi-mono/commit/a74c5da112c29466f182a03108337a488c785d76)to create\n\n[Pi](https://shittycodingagent.ai/), a coding agent.\n\n[, captain of the June team, was standing on Pi’s shoulders on](https://steipete.me/)\n\n**Peter Steinberger**[November 24](https://github.com/openclaw/openclaw/commit/f6dd362d39b8e30bd79ef7560aab9575712ccc11)when he started what became\n\n[OpenClaw](https://openclaw.ai/). That was also the day Anthropic\n\n[released Opus 4.5](https://www.anthropic.com/news/claude-opus-4-5).\n\nOpenClaw\n[exploded](https://www.star-history.com/#codecrafters-io/build-your-own-x&torvalds/linux&openclaw/openclaw&type=date&legend=top-left).\nThe spotlight on Peter is\n[bright](https://x.com/georgepickett/status/2019445652363542911)\n[indeed](https://x.com/lexfridman/status/2021785659644453136). He has now\n[joined OpenAI and announced a\nfoundation](https://steipete.me/posts/2026/openclaw).\n\n(For the record: Armin is [a\nfriend](https://x.com/mitsuhiko/status/1836413328702202276). He and I first\ndiscussed agentic coding at Sentry HQ back in October, 2022, when I prototyped\n[Robb Oat](https://archive.ph/XUhTn) and he was already talking about\ngiving AI agents full shell access. I haven’t met Mario or Peter.)\n\nI watched all this from the sidelines because I was busy [making a\ndocumentary](https://www.youtube.com/watch?v=tOn-L3tGKw0). Since releasing that\non January 20, I’ve been catching up on the latest developments in agentic\ncoding and beyond. What does it mean for Open Source? I see talk of [slop\nPRs](https://github.com/orgs/community/discussions/185387) and [outright\nspam](https://simonwillison.net/2025/Dec/26/slop-acts-of-kindness/) and of\ncourse the [death of Open Source](https://arxiv.org/abs/2601.15494)\n([again](https://heathermeeker.com/2023/10/30/rumors-of-the-death-of-open-source-are-greatly-exaggerated/)).\nWhat’s going on here?\n\n## Reeling From a Sip\n\nI figured I’d better taste the Kool-Aid in order to form an opinion, so I dove\ninto Claude Code with Opus 4.5 on a side project. I spent three 12+ hour days\nwith it. I was intoxicated. My family was weirded out.\n[Observing](https://x.com/chadwhitacre/status/2017021563770405198) how\n[expensive](https://x.com/chadwhitacre/status/2017429682757537907) it is, I\ndecided to time-box my usage. I gave myself two more days to ship an MVP of\nthis side project, and then I would shift to writing up a blog post about my\nexperience with agentic coding and my thoughts on its impact on Open Source.\n\nIt weirded me out too, when I unplugged for a long weekend. Something felt off. It was like I had another “person” in my head, sharing my inner monologue—but the “person” was a computer system owned by a budding megacorp. I spent Monday evening and Tuesday with it again, and this time got so bogged down in bugs and felt so gross that I gave up on shipping the MVP in the time allotted.\n\nI took a step back to reflect on what I had just experienced. That was the day\nPeter’s [SF pilgrimage](https://x.com/steipete/status/2017790488666116514)\nerupted into the [first](https://x.com/francedot/status/2017271738401534310),\n[white-hot](https://x.com/georgepickett/status/2019445652363542911)\n[ClawCon](https://luma.com/moltbot-sf-show-tell). The pull was so strong I felt\nit even in Pittsburgh. I fled the agentic vortex. How far would I have to go to\nclear my head and gather my true thoughts?\n\nI drove north through [a deadly\nsnowstorm](https://triblive.com/local/regional/1-dead-in-18-vehicle-crash-on-i-79/)\non Friday, spent the night at my inlaws’ empty house (they’re in Florida), and\nin the morning sought out an old Amish friend. I beached the car in three snow\ndrifts on the way. Three times Amish bailed me out. The last tried hitching\nhis horse to my car.\n\n## Aaron’s Table\n\nI abandoned the car and walked the last mile, wind whipping deep drifts and deeper chill. Subzero. Stark sun. My reward: spelt raisin cookies and peanuts from the field outside, at Aaron’s kitchen table. “Tell me, what is the World Wide Web? How would you explain it to a cabbage farmer? I’ve been studying up a lot lately. A web is a snare, isn’t it? Why can’t I seem to escape it, even out here? What is money anymore except a number in cyberspace?”\n\nI suggested that perhaps the cabbage farmer ought to be telling me about the Web. Two hours passed quickly.\n\nI left with many gifts, least among them three books. Two are mine to keep:\n*The Complete Works of Menno Simon*, and *Hope Thou In God: Help for\nDepression From the Psalms and Other Scriptures (Study Questions Included)*.\nYes, the Amish are not unfamiliar with depression and mental illness. The third\nbook is to be returned, a zine from the 1990s called *Why We Live Simply*.\nFrom pages 23–24:\n\nThere is no reason to think that the process of change has reached its peak. The snowball effect is sure to continue—every discovery and invention opens the door to dozens more. If the world stands, there will almost certainly be greater changes ahead in the next fifty years than we have seen during the past one hundred and fifty years. There is no reason to think that any end is in sight until the arrival of The End.\n\nI spent Saturday night and Sunday re-engaging with society (the Super Bowl\n“helped”). On Monday I started drafting this post from my local library,\nhard-won clarity in hand. Long story short, **I’ve decided to dial back my\nengagement with mainstream technology, and to launch a print magazine called\nGift** to network with like-minded individuals.\n\nI see two trajectories in technology, either of which alone is off-putting. Considered together, it’s enough for me to draw the line. What are the trends, and what is the line?\n\n## The Attachment Iceberg\n\nMy friend Aza Raskin cofounded the [Center for Humane\nTechnology](https://www.humanetech.com/) (CHT) with Tristan Harris. The organization\n[began](https://www.humanetech.com/impact-and-story) when Tristan, working at\nGoogle, decided to act on his observation of the “detrimental effects of\nattention-harvesting design” in social media. They now have AI in view as well.\nWhat they see is [a crashing wave of attachment\ndisorders](https://centerforhumanetechnology.substack.com/p/the-attachment-economy-is-here-were),\ndue to AI agents “designed to exploit our most fundamental\npsychological vulnerabilities [at] an unprecedented scale.”\n\nWe’ve been here before. Social media was our first mass experiment with AI and it created the\n\nattentioneconomy, leaving us with a loneliness epidemic, rising political polarization, and fractured attention spans. Now we’re running the same experiment with something far more dangerous: AI companions able to hack theattachmentsystem that shapes our identity and bonds us to others.\n\nPeter, Mario, and Armin—leaders of the charge—have all shared their personal\nexperiences with the psychological downsides of agentic AI. “If there’s\nanything I can read out of the insane stream of messages I get, it’s that AI\npsychosis is a thing and needs to be taken serious,” [said\nPeter](https://x.com/steipete/status/2018035483960369161). In\n“[Just One More\nPrompt](https://steipete.me/posts/just-one-more-prompt),” he wrote, “AI was supposed to save time, yet I work more than ever before, I have more\nFOMO than ever before. I dedicate pretty much all my waking time to this, yet\nit doesn’t feel enough.”\n\nIf there’s anything I can read out of the insane stream of messages I get, it’s that AI psychosis is a thing and needs to be taken serious.\n\n— Peter Steinberger 🦞 (@steipete)[February 1, 2026]\n\nMario [chimed in](https://x.com/badlogicgames/status/2020275522182217887), “The\nwinners will be those, who can sustain a healthy life style and not give in to\nthe slot machine unconditionally.” But Armin opened up about just how difficult\nthis is, in a section of [his 2025\nrecap](https://lucumr.pocoo.org/2025/12/22/a-year-of-vibes/) called “The\nMachine And Me”: “I increasingly find it harder not to create parasocial bonds\nwith some of the tools I use. I find this odd and discomforting.” He develops\nthe theme much further in “[Agent Psychosis: Are We Going\nInsane?](https://lucumr.pocoo.org/2026/1/18/agent-psychosis/)” The conclusion\nis chilling:\n\nMaybe this is just the awkward transition phase before we figure out new norms.\n\nOr maybe some of us are\n\ngenuinely losing the plot, and we won’t know which camp we’re in until we look back. All I know is that when I watch someone at 3am, running their tenth parallel agent session, telling me they’ve never been more productive — in that moment I don’t see productivity. I see someone who might need to step away from the machine for a bit. And I wonder how often that someone is me.\n\nSome developers, like [Salvatore Sanfilippo](https://antirez.com/news/158) and\n[Mitchell Hashimoto](https://mitchellh.com/writing/my-ai-adoption-journey),\nseem unperturbed by the psychological rip tide. I’m not as strong. After a few\ndays locked in a productive embrace with AI agents, I resonate with the\nwarnings of the dangers of disordered attachment. It feels like a powerful\nsynthetic opioid binding to my attachment receptors, or something.\n\n## The Machine Starts\n\nThe use of the word “machine” in reflections on AI is telling. While the tech\nindustry was molting OpenClaw, my daughter and I read * Against the\nMachine*,\na Christmas gift from a friend, a grumpy rant against headlong technological\nacceleration. The word is also used in\n\n*The Matrix*. Both hearken to E.M. Forster’s classic novella,\n\n*, in which the mass of humanity live encased in cells, fed through a tube, evacuating through another, while handheld screens permit them to discuss “ideas” with one another. The few outside the Machine are Homeless.*\n\n[The Machine Stops](https://www.cs.ucdavis.edu/~koehl/Teaching/ECS188/PDF_files/Machine_stops.pdf)We’re on our way.\n\nCreeping vines [seem to have\nagency](https://www.youtube.com/watch?v=jlqqZZ7q4-s) in timelapse. I imagine\ntechnology in a century-long timelapse, a tentacled monster forcing itself into\nour orifices. Take screens as an example. They went from:\n\n- curtains hung in public theaters, to\n- massive CRTs sitting on the floor at home, to\n- flat-panel LCDs on our wall, to\n- laptops on our lap, to\n- smartphones in our hand, to\n- smart glasses on our face, to\n- VR goggles covering our face, and now\n- Neuralink is preparing to\n[bypass our eyeballs entirely](https://neuralink.com/trials/visual-prosthesis/).\n\nI wrote a song with my friend Kirk Botula about this progression:\n\nThis second trajectory—tech invading our bodies, embalming us into\nmachine-dependent cyborgs—dovetails with the first. It’s hard to imagine\nagentic AI *not* finding its way into our brains, a constant companion\nthroughout life. “I operate as one half of [a human-AI\ndyad](https://x.com/i_need_api_key/status/2017311428735045852) with my human.”\n\nYeah, no thanks. Hypercapitalism’s *psychological* invasion is bad enough.\nSeeing it in tandem with the equally relentless *physical* invasion motivates\nme to draw a line. No AI agents for me, if I can help it.\n\n## My Amish Moment\n\nThe CHT piece [tries to be constructive](https://centerforhumanetechnology.substack.com/i/186036567/theres-a-better-way-forward):\n\n[T]his isn’t an anti-technology argument. The goal isn’t to eliminate AI from education, therapy, or social connection. It’s to design AI systems that enhance human relationships rather than replace them.\n\n[Y]es, it means AI companions will be less addictive, less profitable, and less “sticky.” But if we want to protect human psychological development, that’s the trade-off we need to make.\n\n[…]\n\nIf we act now, with better design principles, independent research, and a clear understanding of what we’re protecting, we can build AI systems that strengthen human bonds instead of replacing them.\n\nI am settling into the conviction that “we” will not make the necessary\ntrade-off, we will not “act now,” and therefore we will build inhumane AI\nsystems. So while there are still Open Source deck chairs to rearrange—slop\nPRs (Mitchell is [on it](https://github.com/mitchellh/vouch)), perennial\nsustainability challenges (finally getting solved with\n[Pledge](https://opensourcepledge.com/) and\n[Endowment](https://endowment.dev/))—I am now also committing myself to\ndisembarking from the titantic of technological accelerationism.\n\nAll efforts to address the problems of invasive technology are worthwhile, even those that are only partially effective. For my part, I have started trying to return more fully to a pre-screen, analog life. What does this mean? Consider four levels of increasing human autonomy:\n\n**The Baby Step**—Leave your phone in the kitchen on Sunday night instead of sleeping with it. Wake up without it on Monday morning.—Detach from the machine all day, once a week. My friend Julian Cisneros is growing this movement.[Analog Sunday](https://www.analogsunday.co/)**Screen Time/Bean Time**—Silly name my daughter and I came up with for using screens professionally, detaching in private life. Unplugged outside of work/school. Calls but no texting. It gets real because we have to go back to old ways of communicating, finding information, navigating, accounting, etc.**This is where I’m at today.****Analog Economy**—Some avoid the machine in their professional life as well. I’m not there and see no clear path, but …\n\n### Introducing *Gift* Magazine\n\nI’m launching a print magazine called *Gift* to contribute to the conversation\nabout escaping the machine. I intend for the first issue to be an essay laying\nout my vision for it (fair warning: God will be acknowledged).\n\nTo receive a copy of the inaugural issue, write to:\n\nGift\n\nP.O. Box 200\n\nSewickley, PA\n\n15143\n\nUSA\n\nSuggested reciprocation is $10 cash or check (or rough equivalent in your local currency). Be sure to include a return address or otherwise specify where to mail it.\n\nI don’t expect for *Gift* to have an online presence, and I’ll go back to\nshuffling Open Source deck chairs … but in my off-hours I hope to become\nincreasingly Homeless.", "url": "https://wpnews.pro/news/spitting-out-the-agentic-kool-aid", "canonical_source": "https://openpath.quest/2026/spitting-out-the-agentic-kool-aid/", "published_at": "2026-05-29 23:32:13+00:00", "updated_at": "2026-05-29 23:45:10.296685+00:00", "lang": "en", "topics": ["ai-agents", "ai-startups", "ai-products", "ai-tools", "large-language-models"], "entities": ["Sentry", "Claude", "Anthropic", "OpenAI", "Armin Ronacher", "Mario Zechner", "Peter Steinberger", "VibeTunnel"], "alternates": {"html": "https://wpnews.pro/news/spitting-out-the-agentic-kool-aid", "markdown": "https://wpnews.pro/news/spitting-out-the-agentic-kool-aid.md", "text": "https://wpnews.pro/news/spitting-out-the-agentic-kool-aid.txt", "jsonld": "https://wpnews.pro/news/spitting-out-the-agentic-kool-aid.jsonld"}}