Skip to content
The Agentic Review

Independent analysis of agentic AI systems, products, and standards.

Vol. I · No. 1 · Est. 2026
SW

Technical reviewer

Dr. Saul Wenmiller

Saul Wenmiller is the pen name of a working systems engineer who has spent the last decade building distributed orchestration platforms inside two large infrastructure companies. He writes the Review's deep-dives on agent runtimes, scheduler internals, and the protocols emerging around the MCP and A2A working groups. The pen name is by his own preference: his current employer does not permit named bylines on outside publications.

His writing focuses on the unglamorous parts of agentic systems — the queue topology, the retry semantics, the failure modes that surface only at long horizons. He reads source. He runs the systems he writes about. He is patient with vendor claims and aggressive with his own. The Review's editorial policy is largely his.

Saul is reachable via the editors' desk. He does not take direct outreach; pitches to the standards desk should go through editors at agentic dot review.

Joined

Pieces published

Contact

Covers

architecturestandardsMCPA2Ainteropagentic operating systemsAI operating systems

Pieces by Dr. Saul Wenmiller

Standards

MCP and A2A — One Year Under the Linux Foundation

Anthropic donated MCP. Google donated A2A. The substrate of agentic AI now lives under a neutrally-governed foundation. A reading of what that means architecturally, what it means politically, and what the next twelve months of standards work probably looks like.

MCPA2A
Architecture

The Three Metaphors for Agentic Systems — OS, Mesh, Workforce

Three architectural framings are competing for the same agentic-systems problem. Each gets something right. Each papers over something important. A reading of the OS metaphor, Nate Jones's Mesh thesis, and the Workforce model that Anthropic and Sierra have leaned into.

agentic OSagentic mesh
Architecture

Identity in Agentic Systems: The Hardest Unsolved Problem

Identity is the problem the agentic field has spent the least time on and the one that determines whether the systems are deployable in any regulated environment. A working framework for what identity actually has to do.

identityauthentication
Opinion

Why 'Operating System' Is the Right Metaphor for Agentic Stacks

The agentic-OS label gets diluted because it sounds important. It sounds important because it is. An argument for keeping the metaphor and what it actually demands.

agentic operating systemsAI operating systems
Protocols

MCP, A2A, and the Coming Agentic Interop Layer

Two protocol efforts are quietly defining how agents will talk to each other and to the rest of the world. A reading of what they are, what they leave open, and where the lines are still being drawn.

MCPA2A
Cornerstone

What Is an Agentic Operating System, Really?

Most products labeled 'agentic OS' are orchestration libraries with a marketing budget. A working definition, drawn from the parts of the stack that actually have to behave like an operating system.

agentic operating systemsAI operating systems