Many companies are stressed about token cost, as we are moving to a consumption based charge. In the meantime - new open source models, such as DeepSeek V4 Flash and GLM5.2 reduced the price to about 13x chepaer. Also OpenAI had reduced its price for considerably. Now Meta is back in this game. The upcoming months are going to be interesting (GoT)...

Causation or correlation?

I usually learn compelx topics using the image generator of an LLM, capabilities like GPT-Image2 or Nano-Banana. I give it like a complex paper -> turn into visualzation or a poster, then ask quesitons and it helps me understand someitmes a very complex paper rather easily. The jump that nanobanana / gpt-image-2 had done is pretty wild.

i wonder how agents would adopt a whole new programming language, as not much syntax examples / references are out there.

Bit surprised nobody's mentioned the security side. The V8 isolate guards the wrong half of this imo. It stops the agent's code from escaping the sandbox, but the risky part of a browser agent isn't code escaping, it's that it reads untrusted pages and then acts on them. A prompt injection on a page doesn't need to break out of V8, it just tells the agent to use its normal tools to do something you didn't intend. So the isolate bounds what it can run, not what it can be talked into doing. How does Kitesurf scope that part, ie what can the agent actually reach and send after it loads a hostile page?

Right — the isolate protects the host from the agent, but nothing protects the agent from the page. A commerce agent reading 40 product pages is reading 40 untrusted inputs, any of which can carry instructions. Hidden instructions in product descriptions will become the new blackhat SEO: instead of keyword-stuffing for crawlers, stores get prompt-stuffed for agents.