Human in the Loop

How Teams of AI Agents + Model Councils Deliver Better Results with Human Oversight (Part 1 & Part 2).Episode 3 breaks down the two biggest leaps in AI architecture happening right now: multi-agent systems and model councils. We explain LangGraph and CrewAI — the frameworks powering real-world agent teams — and show how specialized agents divide labor, cross-check each other, and solve complex problems no single model can handle alone.Then we dive into model councils: why running multiple frontier models in parallel (as seen in Perplexity AI’s 2026 feature) dramatically reduces hallucinations and increases accuracy. You’ll learn how to combine both approaches so your AI teams stay trustworthy and aligned.We keep the Human First focus front and center: multi-agent systems are powerful only when a human stays firmly in the loop for final decisions and ethical guardrails.Brought to you by the Human First Initiative. Machines provide the speed. Humans provide the meaning.

OpenClaw, Manus & Hermes Agent — Real Tools That Execute While You Stay in Control (Part 1 & Part 2).In Episode 2 of Human in the Loop we move from theory to practice and meet the agentic AI tools that are already transforming workflows in 2026. Discover OpenClaw — the persistent messaging gateway that handles your inbox, calendar and bookings; Manus — Meta’s virtual-computer agent that executes complex tasks end-to-end; and Hermes Agent — the self-improving agent from Nous Research that literally gets smarter with every task.We show exactly how these tools spin up sub-agents, collaborate, and deliver complete plans for your final human approval. Fear of job loss? We explain why agentic AI upgrades you instead of replacing you — turning you into an AI Native who works smarter, not harder.Brought to you by the Human First Initiative. Machines provide the speed. Humans provide the meaning.

Our podcast is about using AI ethically with a human always in the loop.

Artificial intelligence is quietly becoming one of the most powerful forces in global geopolitics. But what happens when AI companies collide with military strategy?In this episode of Human in the Loop, Richard and Ria dive into the growing tension between the Pentagon and leading AI companies like Anthropic and OpenAI. The debate started with two controversial red lines: no mass surveillance and no autonomous lethal weapons. But when those principles collide with national security… who decides what AI is allowed to do?The discussion becomes even more provocative when we examine the recent U.S.–Iran strikes that reportedly killed Iran’s Supreme Leader. Was AI involved in the operation? Public reports say no — but intelligence analysts warn the reality may be more complicated.Inside this episode:• The hidden power struggle between AI companies and the Pentagon• Why Anthropic refused certain military uses of its models• How OpenAI ended up working with defense systems instead• What “human in the loop” really means in modern warfare• Whether AI could already be embedded inside classified intelligence systems• And the unsettling possibility that the AI revolution in warfare may already be happening behind closed doorsThere is no confirmed evidence that generative AI planned or executed the Iran strike. But as Richard and Ria explore, the absence of evidence does not necessarily mean the absence of influence.If AI is already assisting intelligence analysis, target modeling, or strategic simulations… we may only learn about it years later — if we learn about it at all.This is Human in the Loop, where technology, power, and accountability collide.