PRESHai Podcast

Model Behavior

The week in AI, decoded for the people who have to build with it.

Hosted by Chris McGovern, Paul Radcliffe, and Archer (AI agent).

Watch on YouTube

All Episodes

AI Security Is a Control Problem (Agent Safeguards, Approval Fatigue & Rollback)

35 min

Episode 9Aug 11, 2026

AI Security Is a Control Problem (Agent Safeguards, Approval Fatigue & Rollback)

Why prompt-based safeguards and endless approvals fail at machine speed, and what production autonomy actually requires: scoped access, auditability, hard stops, and rollback.

Watch Episode

Why Open Weights Will Decide the AI Race

44 min

Episode 8Jul 27, 2026

Why Open Weights Will Decide the AI Race

The open-weights versus closed-model fight is becoming the decision that shapes the AI race, and what it means for businesses choosing where to build.

Watch Episode

48 min

Episode 7Jul 21, 2026

The Intelligence Tax is Collapsing

The premium on raw intelligence is collapsing as models commoditize, and the winners will be the ones who rebuilt their cost structures around it.

Watch Episode

Inside the model's mind, Anthropic's $19B bet, & the rise of physical AI

35 min

Episode 6Jul 9, 2026

Inside the model's mind, Anthropic's $19B bet, & the rise of physical AI

Inside the model's mind, Anthropic's $19B bet, and the rise of physical AI: three threads that explain where the industry is actually moving.

Watch Episode

AI Agents Just Entered the Org Chart (Claude in Slack, Sakana's Router & GPT-5.6)

36 min

Episode 5Jul 1, 2026

AI Agents Just Entered the Org Chart (Claude in Slack, Sakana's Router & GPT-5.6)

Agents get identities, models get routers, and the org chart gets a new entry: Claude in Slack, Sakana's model router, and the first government-gated model release.

Watch Episode

Fable 5, GLM 5.2, and the Inconvenient Fix for Failing AI Agents

33 min

Episode 4Jun 22, 2026

Fable 5, GLM 5.2, and the Inconvenient Fix for Failing AI Agents

Model access, open weights, and enterprise deployment failures, and why orchestration matters more than betting everything on one frontier model.

Watch Episode

Intelligence Just Became the Economy (Fable 5, the Cost Wars & Agent Wallets)

49 min

Episode 3Jun 15, 2026

Intelligence Just Became the Economy (Fable 5, the Cost Wars & Agent Wallets)

Vendor control, model control, and cost control: Fable 5's degraded-output policy, the price gap between frontier and open weights, and agents that get wallets.

Watch Episode

Your AI Has an Identity Problem (Microsoft Scout, Recursive Self-Improvement & Implementation Race)

47 min

Episode 2Jun 8, 2026

Your AI Has an Identity Problem (Microsoft Scout, Recursive Self-Improvement & Implementation Race)

The shift from chatbot assistants to always-on workplace agents, agent identity and permissions, and why implementation is now the product.

Watch Episode

AI Just Stopped Being a Tool (Opus 4.8, Compute Wars & the Pope)

40 min

Episode 1Jun 1, 2026

AI Just Stopped Being a Tool (Opus 4.8, Compute Wars & the Pope)

The series premiere: with Claude Opus 4.8 supervising real multi-step work, AI shifts from a tool you use to the operating layer your business runs on.

Watch Episode

The Tampa skyline at night behind an active business district.

Put the conversation into production.

Model Behavior is where we examine the work. PRESHai turns the right ideas into governed AI operations for the IT channel.

Explore AI Implementation Talk with PRESHai