Adoption is not competence.
A confidence score is not a policy owner.
A green run is not proof the work happened.
That's the kind of gap I work on.
What do you currently use as evidence that an agent made a better decision, not just a faster one?
https://t.co/iEtLpYRrAd
@PeterDiamandis Moonshot: people+AI deciding better together than either does alone (proof that it's happening)
Who: anyone making big calls with AI in the loop
What changes: avoidable mistakes avoided
Breakthrough: AI collaboration that we can measure, teach, and govern
https://t.co/iHEEpA3KZD
Everybody's favorite @LEJ_limited made another guest appearance on Signals & Subtractions today! In this episode, we talk about some of his most recent open tools and how he compounded his learning to build them, and some other very nerdy things that you can hopefully find useful (or at least inspiring!) https://t.co/sT34wGz1tf
It was good to get my geek on with @LEJ_limited, and he’s never short on opinions either. Developers: how often do you use the Claude Code CLI these days?
Full episode of Signals & Subtractions out tomorrow
@PeterDiamandis If you want to be lazy, AI can amplify that. If you want to be an asshole, AI can amplify that. If you want to be productive and generous and make the world around us better than it's ever been, AI can amplify that. It's about what we direct our Aggregated Intelligence toward.
@EMostaque@jaminball@gradypb And that the only existing competency measures for the human competency side for AI are not well-known yet. Adoption is just the beginning of the trust that's needed between people and agents.
@EMostaque "A race to the top needs a public measure of which way is up." Agreed, and the hard part isn't certifying the model. It's measuring whether the decisions it touches actually got wiser, not just faster. That's the gap I'm working on.
Had an amazing time at #MoonshotsLive on Friday in LA. I thought I might come out of it with a different moonshot for myself, but though I'm totally rethinking my business model, the moonshot & MTP are holding up "To improve how different intelligences work together, so our decisions get wiser, not just smarter."
https://t.co/arNE14g3xx
Critical foundational concepts that Change Management as a field tends to ignore. "What are your people able to accept?" More than that won't be accepted, no matter who sponsors the change or how great it may be!
In this 30min episode, you'll hear:
1. Why it only took 4 days to replace his Hubspot, and Ankit's own caveat on that number. The CLIs, keys, and hosting were already his. You can't go zero to sixty on a bumpy road.
2. That time he gave everyone on his team AI, then took it all back, and rebuilt around learning theory instead.
3. His first question about any task: is it deterministic, or can it be mapped until it is? If yes, it's code, every time. AI writes the program, it doesn't run the business. The worked example is vision insurance pricing, twenty minutes of a person's arithmetic per patient, gone.
https://t.co/eOJV0p9o65
It was great talking with Ankit Patel last night about how he’s used AI to swap out SaaS subscriptions, saving about $2500/m which makes all the difference to the optometry practice he and his wife have. SMBs like his are on both ends of the SaaSpocalypse.
Full episode tomorrow
Sometimes it seems like everybody talking about Agentic AI in #OrgDesign has never run an agent themselves. My guest this week had Kimi and Qwen running while we live streamed, and boy, did it make all the difference in the quality of our conversation!
https://t.co/wpQXqEWibY