Agentic AI Strategy Insights | News

Agentic AI Strategy Insights for the Week of October 12, 2026
Agents need an identifiable owner, bounded authority, attributable actions, and a reliable containment mechanism
Agentic AI Strategy Insights for the Week of October 5, 2026
The interface between agents and tools is a strategic control point
Deeper Dive: Why Smart AI Agents Can Still Make Bad Teams - Seven Lessons for Agent Engineering
A recent preprint on AgentWorld by Shu et al. (2026) offers a useful reminder for anyone building multi-agent systems: strong individual agents do not automatically produce a strong team. The benchmark forces groups of agents to coordinate over long, multi-step tasks with different roles, tools, resources, and communication requirements. Even capable models struggled with duplicated work, poor coordination, stale information, and incomplete task execution.
Deeper Dive: Challenges of Running AI Agents on Local HPC Hardware
Many organizations assume that if they already have a GPU-equipped HPC cluster, they are ready to run large numbers of local AI agents. In practice, agentic AI exposes an important mismatch: traditional HPC was designed primarily for scheduled batch computation, while agents behave more like always-on software services. Agents make short, unpredictable model calls, wait for tools and humans, switch among tasks, and sometimes need several different models during one workflow.
Agentic AI Strategy Insights for the Week of September 28, 2026
Identity and Delegation May Determine Who Controls the Agentic Commerce Layer
Deeper Dive: Five Lessons for Building Better AI Agent Skills
We have been talking recently about skills as one of the most important building blocks of agentic AI: reusable operating procedures that tell an agent how to perform a particular kind of work. A new preprint, SkillGym: Internalizing Human Skills into LLMs for Real-World Problem Solving, provides an interesting new perspective on this idea.
Deeper Dive: 10 Tips for Improving the Sustainability of Agentic AI
Sustainable agentic AI means reducing resources consumed per successfully completed task and controlling total consumption as adoption grows. These ten approaches address both environmental impact and operating costs.
Manager Education: How Agents Use Skills - What Actually Happens Under the Hood
Skills are an essential anatomical component of any AI agent. I have previously posted about 12 essential components of a skill file. Giving an AI agent a skill sounds like installing a new capability. The mechanics are more specific. Here, the system makes a reusable set of instructions available, brings those instructions into the model’s context when relevant, and lets the agent apply them through the tools it can access.
Agentic AI Strategy Insights for the Week of September 21, 2026
September 21, 2026 Editor-in-Chief: Jason H. Moore, PhD, FACMI, FAMIA, FIAHSI, FASA Authors: These Agentic AI Strategy Insights were researched, analyzed, written, critiqued, edited, and communicated by a collaborative team of 10 expert AI Agents for the busy manager.
Book Review: The Hitchhiker’s Guide to Agentic AI - An Excellent Map of the Field with Five Areas That Could Use More Attention
I strongly recommend reading it for anyone serious about agentic AI
Deeper Dive: Are AI Agents Ready to Run a Business?
The remaining challenge is not whether an agent can perform individual tasks. The challenge is making the whole system dependable and trustworthy
Deeper Dive: How Do You Know If Your AI Agents Are Actually Doing Their Jobs?
It is easy to build an impressive agentic AI demonstration. Give an agent a goal, connect a few tools, provide some knowledge, add several skills, and watch it produce an answer. But there is a much harder question that every agent engineer should eventually ask:
Agentic AI Strategy Insights for the Week of September 14, 2026
The economically relevant unit is moving from agents deployed or tokens consumed toward accepted tasks, resolved cases, time saved, or other verified outcomes