A custom AI agent runs faster and cheaper when you strip out what it doesn't need. Strip out too much, and it stops reasoning well. Alon Goren on finding the balance in the latest AI, Actually. Full episode: hubs.li/Q04zhXPh0 #AIActually #AgenticAI

Lines of code and pull requests are easy to count. Are they the right way to measure AI? AnswerRocket co-founder Mike Finley has a sharper answer on the latest AI, Actually. Full episode: youtu.be/QvGtwrqmavk?si=6ryH… #AIActually #AIStrategy

AnswerRocket, Balfour Beatty, and Microsoft teamed up on a national AI hackathon. At LCI Congress in Atlanta, we're breaking down the framework it produced. Wed Oct 14, 1:00-1:45pm. Register now: hubs.li/Q04yWWv50 #LCICongress #AIInnovation

Most AI tools are built for one user. Enterprise workflows need a whole team. Ep. 26 of AI, Actually: Pete Reilly, Alon Goren, Mike Finley & Stew Chisam on multiplayer AI, harnesses, the AI pause debate & JEV. 🎧: hubs.li/Q04yw7QR0 #AIActually #EnterpriseAI

Not every AI task needs a full LLM. Mike Finley explains JEV, a lighter, faster approach that just answers yes or no instead of running every small decision through a big model. Episode 26 of AI, Actually drops Tuesday, 9/29. #AIActually #EnterpriseAI

Slowing frontier AI won't slow the race. Open source is already global, and better models keep coming no matter what any one country decides. Alon Goren breaks it down on the new AI, Actually clip. Watch here: hubs.li/Q04y2f2p0 #AIActually #AIRegulation

Usage-based pricing makes sense for accessing a model. It doesn't necessarily make sense for the AI solution built on top of it. New episode of AI, Actually on what enterprises should really be pricing. 🎧 youtu.be/dh-xdBS2uOs?si=qIqP… #AIActually #EnterpriseAI

Never let an agent evaluate itself. Ada Gil on why the real challenge in enterprise AI isn't defining the outcome, it's defining "done" before you start measuring it. hubs.li/Q04xKyMn0 #AIActually #AgenticAI #AIGovernance

The credential leak problem in AI agents usually comes down to one design flaw: generated code and secrets sharing a container. Here's how Claude Managed Agents' brain/hands/session architecture closes that gap: hubs.li/Q04x6XVs0 #AgenticAI #AIArchitecture #EnterpriseAI

AI agents don't need to start at senior level. Start them as interns and promote them, same as any new hire. New episode of AI, Actually is live. 🎧 youtu.be/dh-xdBS2uOs?si=7yAY… #AIActually #AgenticAI #EnterpriseAI