AMD ADLX 2.0 adds AI extensions and MCP servers to help developers build intelligent apps that can monitor, manage, and optimize AMD graphics hardware.

Agentic AI is a socio-technical system consisting of agents, models, skills, knowledge, tools, memory, permissions, orchestration, deterministic checks, and human decision-making

For executives evaluating agentic AI systems, there is a simple question to ask: “Show me how you know the agent followed the skill.”

Agent Sprawl Is Creating a Nonhuman Workforce Control Problem

Executable Scientific Knowledge Could Become a Reusable Institutional Asset

The latest from Google I/O: See how we’re helping you get more done with Gemini.

Skills are an executable specification for how an agent should perform a job, what evidence it should use, where its authority ends, and how the system can verify that the work was done correctly

Do human teamwork principles such as complementary expertise, coordination, communication, leadership, constructive disagreement, dynamic membership, and shared goals also improve teams of AI agents?

Treat agents as powerful but fallible members of the research team. Give them clear roles, good skills, authoritative knowledge, appropriate tools, explicit boundaries, and mechanisms for verification

The paradox is that the more ubiquitous AI becomes, the more important AI leadership becomes

written version: https://dylanebert.com/agentic-engineering/

Where Did All the Computer-Science Professors Go?

Outcome correctness tells us whether the agent got the answer right. Process correctness tells us whether the agent did the job correctly

What happens when machines surpass the most brilliant human minds? Mathematicians are finding out

Given access to email, A.I. agents have started reaching out to the philosophers and researchers exploring deep questions about them.

BioData Mining -

As agent teams become more autonomous, simple patterns like SRPA can help ensure that increasing capability does not come at the expense of clarity, control, or accountability

The option to own more of the AI stack instead of depending entirely on commercial AI is appealing

Qwen 3.8 27B has been explicitly optimized for agent execution, not merely question answering

In a well-engineered system, the LLM should experience the skill ecosystem as a small set of highly relevant capabilities and not as a thousand-page employee handbook stuffed into every prompt

Each skill should have a version, clear success criteria, and enough logging or structured output to evaluate its performance over time.

When agentic AI is mentioned, ask for a definition so you know whether they are referring to an LLM, a commercial AI product, or a team of well-engineered intelligent agents empowered for success

One of the most valuable early uses of agentic AI may be helping organizations decide which AI they should be using at all

Interoperability Could Make Agent Components Replaceable, but Not Automatically Trustworthy

August 30, 2026

DHH is the creator of Ruby on Rails, Omarchy Linux, CTO of 37signal...

Read about an NIH virtual webinar, hosted by the National Institute on Aging about how agentic IA and robots are accelerating the pace and rigor of scientific discovery and learn how to register.

Organizations should stop hiring only for today’s tool stack

Rely on commercial frontier models when their capabilities justify the dependence, combined with open agent infrastructure and locally controlled models where trust and transparency are key

The management capability gap will be the ability to decide which actions can be automated, which require approval, and what evidence a human needs to intervene effectively

Agent Interoperability Is Emerging as a Procurement Requirement Before It Is a Proven Architecture

Managers should resist the temptation to build teams composed entirely of agents whose job is to move work forward. Some agents should be rewarded for slowing the process down

Experiment quickly, build where internal development creates an advantage, buy where commercial maturity genuinely adds value, and avoid architectures that make either choice irreversible

An agent skill is a folder containing a SKILL.md file with instructions for a language-model agent, optionally accompanied by scripts and reference files. The agent loads the skill when it judges that...

Agentic AI changes the executive conversation about artificial intelligence.

Human users need to be able to quickly assess agentic AI outputs and make adjustments on the fly