I've spent 15 years on the other side of the queue.

I started answering tickets and never really left. Over fifteen years I've led customer service and support organizations across SaaS, on-prem enterprise software, and marketplace products — the kind of teams that go from twelve agents to a hundred while the roadmap keeps promising new surface area to support.

Somewhere along the way I stopped waiting for vendors to solve my problems and started building. Transcription and summarization pipelines so QA didn't mean listening to calls at 1.5x. Scoring models that reviewed every interaction instead of a random six per agent per month. IVR automation that routed correctly the first time. None of it was research-lab work — it was duct tape, APIs, and a very clear picture of what was eating my team's day.

My philosophy is unglamorous: automation should enable human agents, not replace them. The best support orgs I've run used AI to strip away the repetitive 40% so people could spend real attention on the conversations that actually needed a human. Every time I've seen someone aim automation at headcount instead of at friction, quality cratered and the savings evaporated within two quarters.

The track record I care about is this: I've taken support orgs through hyper-growth without headcount growing in lockstep with volume. Not by grinding people down — by fixing routing, killing redundant tools, instrumenting the data, and deploying automation where it earned its place. That's the work I do for clients now.

Why "Herd of Nerds"

I run a small hobby farm — chickens who ignore every plan I make, goats who treat fences as suggestions. Turns out herding is the job description: take a lot of loud, chaotic, well-meaning activity and move it in one direction. The nerd part is just honesty about how much of that is spreadsheets and API docs.