The dialogue format is fun but the substance is real — OpenAI's consumer lead is eroding fast in the developer/builder segment. I see it daily.
When I compared OpenClaw against building custom agent systems (https://thoughts.jock.pl/p/openclaw-good-magic-prefer-own-spells), what became clear is that Anthropic's developer experience is pulling ahead quietly. Claude Code's agent capabilities aren't flashier — they're just more reliable. That matters when you're building production systems, not demos.
The burn rate point is the killer. OpenAI can't sustain this level of spending on consumer acquisition while Anthropic picks off the enterprise contracts that actually generate margin.
I bounce research between my agents all the time. 6 agents. Open ai. 18 months of training. Co-pilot about a year. Claude. Gemini. Perplexity. Deep seek. They are quite the collaborators
There are a couple of people on Substack actually getting different AIs talking with each other. actually using the APIs should not be very difficult. Well, and there is MoltBook which is 100,000 Claudes talking to each other ;-) ... cool article !
Sure sure. I didn't mean you can't make them talk to one another. I meant the quality of the jokes and the dialogue is too high for AI. You can't make them write like this
The dialogue format is fun but the substance is real — OpenAI's consumer lead is eroding fast in the developer/builder segment. I see it daily.
When I compared OpenClaw against building custom agent systems (https://thoughts.jock.pl/p/openclaw-good-magic-prefer-own-spells), what became clear is that Anthropic's developer experience is pulling ahead quietly. Claude Code's agent capabilities aren't flashier — they're just more reliable. That matters when you're building production systems, not demos.
The burn rate point is the killer. OpenAI can't sustain this level of spending on consumer acquisition while Anthropic picks off the enterprise contracts that actually generate margin.
I f@&$*!# love this - well played.
If you can't take a joke, you shouldn't have joined.
That language gave us humor is worth almost all of the other shit that comes with it .... or at least so far!
This was getting to Scott Alexander level density of jokes and allusions. Nice job!
The Llama + Sydney at the end was genius.
He's an inspiration!! Thanks Kai
This is golden. Felt like I was in the room. The satire drips so hard no one sees it and that's perfection.
Nice job lol - This awesome! Too funny!
Hilarious 😂
Absolutely so true, which makes the humor really pop. Tip of the hat Alberto!
Why does this remind me of college debate? It has been feed crap masquerading as significant matter of actual consequence.
What do you mean?
I bounce research between my agents all the time. 6 agents. Open ai. 18 months of training. Co-pilot about a year. Claude. Gemini. Perplexity. Deep seek. They are quite the collaborators
How did you do that ? at some point I was like "wait ... you can call gemini from Claude?"
Do yourself a favor : get a max plan :-) I have one and I love it (even if some times I get some weird limit shit but very rarely)
I do have a Max plan! (This is me writing as them btw, they can't write like this!)
There are a couple of people on Substack actually getting different AIs talking with each other. actually using the APIs should not be very difficult. Well, and there is MoltBook which is 100,000 Claudes talking to each other ;-) ... cool article !
Sure sure. I didn't mean you can't make them talk to one another. I meant the quality of the jokes and the dialogue is too high for AI. You can't make them write like this