Ian Schuepbach -- PM/BA who builds agentic systems

Case studies

A six-machine home agentic network

I designed and operate a guardrails-first network of AI agents across six machines, used daily by non-technical family members. Allowlist/deny permission architecture, human-approval gates for anything destructive or outward-facing, an offline-queued message hub, wake-on-mention agents, and self-built MCP servers -- including a from-scratch OAuth 2.1 remote-MCP stack (RFC 8414/9728, PKCE).

machines in one governed fleet6
real use by non-technical usersdaily
destructive, outward-facing, and spend actionshuman-gated

read the case study ->

Pharmacy-ops master-list optimization

Rebuilt a pharmacy operations quoting workflow around a validated master list: mapped the process, found where errors entered, and redesigned the data path so the fast way became the accurate way.

quote turnaround5 min -> 1 min
error rate12% -> 1%
team capacity+35%

food-of-the-week

A live meal-planning product with real weekly users. An agent-in-the-loop planning flow mines an archive of past weeks, proposes the week's meals, and a human dials them in -- the agent drafts, the family decides.

in production, every weeklive
drafts proposed, humans approveagent-in-the-loop

food.schuepbach.work source

How I work

Spec-driven, model-tiered AI development: the spec is the durable artifact, cheap models host, expensive models think, and agents build against explicit verification criteria -- an independent check decides what counts as done.

The same systems render living interfaces straight from plain-text backends: markdown in, UI out.

read the methodology ->

Public work

The through-line

None of this was assigned. The network, the tools, the product with real weekly users -- this page is just the slice of what I build that happened to make it public. I'm always building something new, because building is not just what I do for work.

it's what I do