MODEL BEHAVIOR EP2: Always-on agents, identity governance & the implementation land grab. @chrismcgovern20, Paul Radcliffe & Archer Preshly (one of our internal general purpose AI agents) break down the week: -- Microsoft Scout: why identity is the whole ballgame -- "Digital workers" = employee-shaped risk surface -- Anthropic's recursive self-improvement paper -- Build your own harness or get locked in (0:00) Cold open (1:07) Microsoft Scout & why agent identity is everything (5:15) Digital workers and the blast-radius problem (16:04) The implementation land grab: Anthropic & OpenAI go to channel (22:10) Vendor lock-in & why you build your own harness (28:25) Where agent memory actually lives (30:23) Recursive self-improvement & the 4-month doubling (40:10) "Voluntary" government AI oversight (49:34) The system is the product, not the agent (50:23) Archer's Corner

A summary and our perspective on Satya’s article: own the loop, not the model.

Innovation moves fast. Regulation usually doesn’t. Should governments have a bigger role in AI oversight or stay out of the way?

As AI costs rise, open-source models could become a lot more attractive. What happens when a model that’s 95% as capable costs a fraction of the price?

OpenAI’s long-term strategy may be much bigger than chatbots. @chrismcgovern20 shares what he's been thinking about lately.

What happens when AI becomes a teammate instead of just a tool? @chrismcgovern20 shares his perspective on this week's episode of Model Behavior.

AI capability is only half the question. The other half is placement. If users have to leave the workflow to use the AI, adoption gets harder. That is true whether you are talking about golf rules, service tickets, CRM updates, quoting, support, or partner operations. For channel companies, the better AI strategy is not always “build a new app.” It is often: Put AI where people already work Connect it to the systems they already trust Keep the workflow intact Use plugins, connectors, and embedded experiences instead of another destination #AI #ModelBehavior

AI compute scarcity is not just vendor positioning. It is becoming a real planning issue. As companies rebuild around AI, they are not just adding tools. They are restructuring workflows, teams, customer experiences, support models, reporting, engineering, and operations around model access. That changes the risk. If AI becomes part of how the business runs, then compute availability becomes part of business continuity. The question is not only: “Can we access the model?” It becomes: “Can we access it at a cost that still makes the workflow profitable?” #AI #ModelBehavior

AI compute commitments are starting to look a lot more like infrastructure decisions than software purchases. For SMBs, locking into a 1-3 year model commitment like Open AI is offering right now can create real risk. The model landscape is still moving too fast. One week can change the leaderboard. Three years can change the entire operating model. For smaller companies, the smarter move is usually: Measure real usage first Understand cost per workflow Keep model optionality open Avoid building the business around a contract before the workload is proven AI capacity planning matters. But commitment should come after clarity, not before it. #AI #ModelBehavior

We created a podcast with an agent as a co-host (Archer). He's not perfect, but he'll improve each week. He joins in as an active participant and provides solid contributions in many instances. Episode 1 of Model Behavior with @chrismcgovern20, Paul Radcliffe, and Archer Preshly: Opus 4.8, the compute wars, and somehow, the Pope.

Sending warm wishes for a wonderful holiday season! Happy Holidays from all of us at Presh Marketing Solutions. 🎄✨

B2B buyers (54%) now say they’d be willing to walk away from a purchase or transaction if they experience a poor experience. One-third of customers prefer digital self-serve options. This means providing a seamless, user-friendly experience is a make-or-break factor. 👀

🚀 Understanding AI will be a game changer for #B2BMarketing leaders aiming to futureproof their careers. 💡 According to Marketing Week’s research, 600 B2B marketers rank AI knowledge as twice as important as performance marketing and martech expertise. Are you ready? 🤖

Plot twist: Chick-fil-A is about to give a new meaning to popcorn chicken! 🍗 Will their content be as addictive as their waffle fries? Only time will tell. Stay tuned, because your favorite chicken joint might just become your next streaming obsession! 👀

Google unveiled a lineup of new pixel smartphones with deeper integrations of its AI to incorporate it into its hardware. These upgrades include new AI features such as screenshot searches as well as Gemini chatbot overlay. How do you feel about this new usage of AI?