Michael Gulden

Meridian, Idaho — Fractional Chief AI Officer · AI Systems Engineer · Founder

I rebuild business processes as agentic systems — and run them in production.

Thirteen years of end-to-end delivery, now focused on AI: I decompose how companies work, rebuild the highest-leverage processes as multi-agent workflows with human-in-the-loop controls, and operate them daily on infrastructure I built. Talk to a human who builds AI.

See the projectsRead the resume

Track record

In technology — web, UX/UI, and end-to-end delivery since 2013
13 years
Agentic systems design, build, and operations
Full-time AI since Nov 2025
Production agent runs baselined in the Prismatic fleet-reliability program
213 runs
Operated daily by the agent fleet this site documents
2 businesses

Featured work

  • Prismatic Engine

    Open-source multi-agent orchestration engine (AGPLv3) that runs a production fleet of AI agents operating two live businesses daily — with mechanical verification on every output.

    • Multi-agent orchestration
    • Python
    • Human-in-the-loop
    • Open source

  • Human Design Engine

    Commercial chart-computation API and MCP server — AI capability packaged as a product that other systems and agents consume programmatically.

    • API design
    • MCP server
    • Product engineering
    • Security hardening

  • BeyondSaaS.ai

    AI implementation consultancy — in-VPC, cloud, and hybrid deployments for companies that need AI working inside their own operations, with their authenticity and data sovereignty intact.

    • Fractional CAIO
    • Process reimagination
    • Workshops
    • Agentic workflows

  • Production AI Infrastructure — Self-Hosted

    The self-hosted infrastructure my AI work runs on: a multi-host Proxmox cluster serving local LLM inference to a production agent fleet and two operating businesses, daily.

    • Proxmox
    • vLLM
    • GPU infrastructure
    • k3s
    • 40GbE / RDMA

How I reimagine a process

  1. Map the current state

    Work with the people who do the work to document how a process actually flows today — effort, risk, and handoffs included.

  2. Eliminate and simplify

    Remove steps that add no value and simplify the rest before any automation is designed. Automating a broken process just makes it fail faster.

  3. Agentify what remains

    Decompose the surviving process into agent tasks with tool and data access, context strategy, and least-privilege permissions per step.

  4. Build with human-in-the-loop controls

    Prototype and build the agentic workflow with approval checkpoints and evaluation criteria, so autonomous behavior is validated before it is trusted.

  5. Measure and codify

    Capture the value baseline, measure at 30 and 90 days after release, and codify what worked as reference flows and playbooks the team keeps.

The short version

I started in web development and UX/UI in 2013, ran projects end to end for clients — discovery, design, build, handoff — and founded two operating companies along the way. Since November 2025 I work full-time in AI engineering: an open-source orchestration engine, an API + MCP product, a fractional CAIO practice, and the self-hosted GPU infrastructure all of it runs on.

More about me

Current work: BeyondSaaS.ai (fractional CAIO practice) ·Prismatic Engine (open source, AGPLv3) ·Human Design Engine (API + MCP) ·github.com/mbgulden

Building AI into your operation?

I take on a small number of fractional CAIO and AI implementation engagements. Email is the fastest way to reach me.

Email — Mbgulden@gmail.comConnect on LinkedIn