- AI with Kyle
- Posts
- AI with Kyle Daily Update 129
AI with Kyle Daily Update 129
Today in AI: RIP Clawdbot, Long Live Moltbot!
Kyle Balmer
January 28, 2026
What’s happening in the world of AI:
Highlights
🦞 Clawdbot → Moltbot: The Name Change
It happened. Anthropic sent a cease and desist. That was quick!
Clawdbot (C-L-A-W-D) is now Moltbot (M-O-L-T). Same lobster, new shell. As Peter Steinberger, the creator, put it: "Anthropic asked us to change our name. Trademark stuff. And honestly, molt fits perfectly - it's what lobsters do to grow."
A lot of people online are furious. "Screw Anthropic, stepping on the little guy." "How is Clawd a trademark violation of Claude?"
Honestly: fair. In IP law, if you don't protect your trademarks, it hurts you later. You can't pick and choose. If you let one infringement slide, that becomes precedent when someone else infringes. This is why Disney, Nintendo, and Microsoft aggressively pursue trademark violations - not because they're evil, but because not doing so dilutes their intellectual property.
It's not Anthropic sitting in their office thinking "let's screw this guy Peter." It's just how IP law works. Sucks still!
Poor Peter has had a rough week: his project went viral, scammers set up fake crypto coins, hackers tried to take his GitHub, and a multi-billion dollar company sent him a cease and desist. If you're seeing this Peter - what you've built is amazing. Keep going!
FYI: The guide on my site still works - I'll update the URLs to redirect from both /clawdbot and /moltbot !
📉 Riley Brown's Prediction
Worth noting this take from Riley Brown:
I agree. What we're seeing is a preview of the next phase of AI. Clawdbot is a proof of concept - a fully functional autonomous AI assistant. But it's still technical to set up and has security concerns. A lot of them!
Someone will take this and make it cleaner, easier, safer. Apple could do this with Siri + Gemini (they have the Google deal now). Google could do it. Any of the frontier labs could look at the AI internet losing its mind over one guy's open source project and think "we could do this properly."
If you feel like you're missing out - don't worry. Wait a bit! A polished version is coming.
📜 Dario Amodei's 15,000 Word Essay
The head of Anthropic published a massive essay on AI risks. I've started reading it - 15,000 words is a lot, Dario. It expands on what he discussed at Davos (covered last week in the newsletter).
The five risks he covers:
1. Autonomy risks - AI gaining its own free will and wanting to do its own thing. The "I'm sorry Dave" scenario.
2. Misuse for destruction - Bad actors using these tools for terrible things. The technology is neutral but not everyone using it is.
3. Misuse for seizing power - Governments using AI to enable autocracy. An arms race between those using AI for control and those fighting it.
4. Player piano (economic disruption) - 50% of white-collar jobs gone in a year or two. Dario elaborated on his Davos comments. He and Demis Hassabis both agreed: even if AI takes us to utopia eventually, the next few years will be extremely messy and horrible for a lot of people.
5. Black seas of infinity (unknown unknowns) - The real challenges are the ones we can't even foresee. Five years from now, ten years from now, things will blindside us that we cannot predict sitting here today. These are the scary ones.
He ends on a positive note: humans are resilient and muddle through most problems. I broadly agree. But that's cold comfort if you're the one losing your job right now.
This is why we need to engage with this technology. The billionaires in San Francisco will do well from AI regardless. We as normal people need to work out how we benefit too.
Source: Dario’s Full Essay
💰 ChatGPT Ads: $60 CPM
OpenAI is drip-feeding information about their ad rollout. Smart strategy - let people get furious, forget, then drip more info. Boiling frogs.
New info from The Information: ChatGPT ads priced at $60 per 1,000 impressions.
For comparison:
Meta ads: ~$8 CPM
YouTube: ~$5 CPM
TikTok: ~$5 CPM
Broadcast TV: ~$45 CPM
Podcasts: ~$21 CPM
If true, ChatGPT ads are premium priced from day one. Surprising!
I previously said "jump on ChatGPT ads when they launch because new platforms are usually cheap." I was wrong if this holds! Either:
This info is false (possible!)
OpenAI know from testing that ChatGPT conversions are so valuable they can charge premium
OpenAI doesn't know what they're doing
We'll find out soon.
🛠️ Andrej Karpathy on Claude Code: Coding vs Building
The ChatGPT co-creator shared notes on using Claude Code. Here’s the full mini-essay:
This quote is important:
"Fun. I didn't anticipate that with agents, programming feels more fun because a lot of the fill-in-the-blanks drudgery is removed and what remains is the creative part. I also feel less blocked/stuck and experience a lot more courage because there's almost always a way to make positive progress."
"I've seen the opposite sentiment too. LLM coding will split up engineers based on those who primarily liked coding and those who primarily liked building."
That last sentence explains the divide we're seeing.
If you love writing syntax line by line - you probably hate AI coding. It might be better than you, it's removing your job, and it's taking away something you found fun.
If you love building things - this is exciting. The drudgery is removed. What remains is creative.
Both reactions are valid! It depends who you are.
We're moving up levels of abstraction. Just like we moved from punch cards to assembly to high-level languages. Andrej Karpathy said it a year ago: "The new coding language is English."
🎓 On Workshops and Monetizing AI
Someone on my Live asked about making money with AI. My answer: workshops.
The most I've charged for one hour of work is $4,000. Sounds like a lot - but there were 400 people in the room. That's $10 per person per hour. For the company, that's cheap.
Even at $2,000 for 50 people, that's $40 per person per hour. Still very reasonable.
The niche opportunity is the intersection of your AI knowledge and your industry knowledge. You don't need a PhD. You need to translate AI to people in your domain.
If you're a dentist, teach dentists why AI matters. If you're a teacher, teach schools. If you're an aeronautics engineer, teach aeronautics engineers. Niching down also helps marketing. "AI workshop facilitator" is too broad. "[Industry] + AI" is marketable.
→ [Free workshop course at aiwithkyle.com/courses]
Member Question: "Do you have to be technical to set up Clawdbot?"
Kyle's response: Yes and no. You need to be okay with terminal. If you've never used it, it looks terrifying - a flashing cursor where you input commands. But realistically you're just copying stuff from my guide, pasting it, pressing enter. If that intimidates you, wait a few months for someone to make a one-click installer. If you're happy to learn, you can do it.
BUT: there are risks. And those risks are exacerbated if you don’t know what’s happening!
Member Question: "If multiple users (in Clawdbot) share an agent, will it differentiate each user?"
Kyle's response: Yes. It knows who's messaging based on WhatsApp number/device. You'd then instruct Clawdbot how to handle different users - "this is my business partner, give him calendar access" etc. You're essentially setting up permissions like a manager giving instructions to an executive assistant.
Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.