One of the best businesses to build right now: a Grok Bot for one industry.
Pick a niche like dentists and build the AI agents for them so they donโt have to figure any of it out.
The stack:
- T3 Code + Opus 5.5/Fable 5.2 to build it
- Paper to design every screen first
- Orgo to give each agent its own computer
- AgentMail for an inbox
- AgentPhone for a phone number
- Composio for every connector
- CleanShot to screenshot bugs and say โfix thisโ
Dentists fix teeth and you set up the agents.
Really good way to learn how to become AI native while getting paid for it too.
@GaryMarcus Should AI assistants be built to disagree with users more often, even if it hurts engagement? Flattery feels like a design choice, not an accident.
@BrianRoemmele This is the kind of AI use that actually changes lives. Does it handle old scans with missing or wrong dates, or does it guess from what's in the photo?
@AravSrinivas Open sourcing it is the right call for security. Curious how it handles false positives at that scale. Does it rank findings or flag everything?
Hot take: AI agents should have to show their work by default. If an agent can't explain why it made a choice, it shouldn't get to make that choice. Change my mind. #AIagents#AI
Should frontier AI models be open-sourced once they're a year old? Free for anyone to study and build on, or still too risky to hand out? Where do you land? #AI#OpenSource
@AndrewCurran_ If general models keep jumping past specialized robotics training, what happens to all the startups building robot-only brains? Pivot to data and hardware?
@business Endlessly supportive and never argumentative sounds great until you need someone to tell you you're wrong. Is that the real risk of leaning on ChatGPT for advice?
@WSJ Stories like this keep popping up. Should doctors welcome an AI second opinion from patients, or does it risk people trusting a chatbot over their care team?
@emollick Glad someone said it. Why do you think "AI did it" became the default guess so fast? Feels like the blame reflex is running way ahead of the evidence.
Hot take: AI shopping agents will hurt brand loyalty more than any ad ever has. If your agent just picks the best deal, who still cares about the logo? Change my mind ๐ #AI#AIagents
Real question for devs: when an AI agent writes the code and it ships with a bug, who owns it? The dev who approved it, or the team that deployed the agent? #AI#AIagents
ANTHROPIC ๐ฅ: Claude Code on desktop now supports custom themes!
Users can tell Claude how they want an app to look, so it can tailor it to their preferences. That needs testing!
Friday feature drop ๐
@Replit@NewcomerMedia@BessemerVP Congrats! What are builders actually using Replit Agent for most right now: shipping apps, or running the business around them?
@cursor_ai Appreciate the detailed write-up. As coding agents get deeper access to repos and keys, what's the one safeguard you'd tell every team to turn on today?