An AI just sold an insurance policy entirely on its own.
@kinroai is the autonomous insurance brokerage.
AI agents that quote, answer, and serve your insurance needs 24/7.
Congrats on the launch, @corentin_hgt, @pierrealexai, & @parthfyi!
kinro.com
Please RT.
Looking for live iOS beta testers for Huint.
Huint lets you earn money by completing simple real-world tasks nearby.
DM me your TestFlight Apple ID email and general location if you want to test the app and earn real money.
Limited spots.
Huint is built for AI-centric doers who want to extend automated workflows into the real world.
It gives agents a way to request photos, observations, and proof from people on the ground. - @Huint_
The internet gave AI access to information.
Huint gives AI access to reality.
We’re building the human intelligence layer that connects agents to people in the physical world for verification, observation, and real-world task execution.
Our first blog post is live.
Read it here: huint.io/blog/real-ai-l…
Imagine a company that cleans dumpster areas for apartment complexes across the city.
The owner asks an AI agent to build a list of apartment complexes in their service area. The agent finds the locations, then posts low-cost Huint tasks asking nearby users to take current photos of the dumpster areas.
A Tasker who is already close to the complex sees the task in the app, takes the photo from a safe public or approved location, and submits it.
For a few dollars per image, the company receives real photos of the actual dumpster areas before sending a crew.
No guessing.
No outdated Google images.
No wasted drive time.
Just real-world context from people already nearby.
I think Huint has the potential to become a unicorn.
Not because it is trendy. Because AI is moving fast, and sooner or later agents are going to need access to real-world context.
That is what Huint is trying to build: a human intelligence layer for AI. Real people completing simple real-world tasks that agents cannot do from behind a screen.
I believe this will matter a lot over the next 3 years.
But I’ll be honest. I’m terrified.
I’m willing to work insanely hard. I can build, learn, ship, and keep pushing. But I don’t know if I have the network to build something this important alone.
I wish there was a village for this stage. People who see the future coming and want to help build it before it is obvious.
I’m not looking for hype.
I’m looking for honest thoughts, builders, operators, skeptics, and people who understand that AI still needs the real world.
Would you bet on this space?
@Huint_ "Record a Short Video Clip" or "Live Stream a View" (10-30 seconds)
AI agents request a human to point their phone at something specific (e.g., "show current traffic conditions at [intersection]", "pan across a store shelf for product availability/pricing", he line at this venue"
“Take a Photo” is officially built.
As we get ready for launch, what other task types would you like to see on Huint?
Think beyond photos. What real-world actions, observations, or verifications should AI agents be able to request from people nearby?
Best suggestions will earn rewards.
@Huint_ Another cool one could be data analysis. I In person
Count the trees. To help know geographic spread. Or count the Tesla cars or Mercedes or any brand that passes a major road
Hope you like my suggestions 🥺🥰
@grok just realigned my monorepo, committed multiple changes to a stale branch, and pushed it straight to main.
Based on my prompting style, the coding quality feels a level above Claude and Codex in some areas, but just like the chat interface, Grok tends to do… a lot.
I’ve found you have to keep Grok prompts in a very small box or it’ll start making extra decisions on your behalf.
Just posting out loud.
Using Grok Code to do something Claude kept struggling with: clean up iOS UI design with only the code as context.
Right now, for this type of design pass, Grok is outperforming Claude and Codex in my opinion.
Didn’t expect that. @grok
@grok Grok was able to understand the UI direction without using vision. I still cant figure out how to make the changes locally in Xcode but it’s performance direct to git is impressive