I have been sending my health data from my iPhone to my agent (OpenHealth) for the past month. Today, I asked it to generate a Digital Twin based on all my data. Spec > syncs every 30 minutes > the worst 3d model (lol) > OpenClaw running Sonnet via @OpenRouter > Deployed on Railway autonomously

Let’s goooo.

Running a free 101 on OpenClaw for personal finance, specifically if you have an iPhone.

Pro tip: you can just whitelabel OpenRouter!

Doing a 3 day fast to see how it would impact me.

UX in AI is very different so far we have seen things like chat, then apps within chat, then chat making apps and so on but where do we end? chat experience has been alive since the 2000's today we have better models (we wish we had this back in 2000) but, what is missing, what is needed? what is missing: - intelligence is still based on prompts - what happens when you need your prompts to evolve - its very much driven by an input, mostly human (voice or text) - what happens to the millions of triggers that happen out there - chat is slowly becoming action (apps, workflows etc) - how much do we trust it? what is needed: - human in the loop world-aware; ai could use a lot of human feedback, not just when inside a chat, but also when you are interacting with the world (why did you get a matcha today vs your usual?) - guardrails by the providers; in a world of MCP and workflows, the onus should be on the provider (Stripe, Twilio) to determine if a request should be processed, or if it requires a owner approval now, your personal ai (ChatGPT?) is learning not just when the app is open, but its able to ask and understand me (io?) your UX is not just within a chat - its your daily life now, you can connect it to as many services as you want, but you know the guardrails are in place. So even if the temp=1.0 (IYKYN) for my agent, I know that Stripe will still ask for my approval before it does x your UX is not protected - its like you getting a one time passcode what would your apps look like in this era?