MMichael Z Consulting
about

A real person, building for real businesses.

I'm Michael. I'm based in Vancouver. I run a small business of my own — which is the entire reason this practice exists.

/ where I'm coming from

Running a small business taught me one specific thing very quickly: most of what eats your week isn't hard, just repetitive. Quoting, scheduling, follow-up, reporting, hiring, chasing payments. Each one feels small in isolation; but the real cost isn't the time itself — it's the leads you don't notice losing, the mental weight that compounds, and the business that quietly breaks the moment you step away. “Save 10 hours a week” is the wrong framing — most owners can't name what they'd do with the hours. The right framing is the loss you can't see.

The strange part was looking around for help. The AI automation consulting market is full of people selling 30-minute discovery calls and abstract strategy decks. I've sat through a few of those. I wanted someone who would just look at my real workflows, tell me what to build, and then actually build it or honestly tell me I could do it myself.

That gap is why I started this. I build for other SMB owners the way I'd want someone to build for me.

/ how I work

No agency, no SaaS product, no monthly retainer by default. I pick tools per project — n8n, Make, Zapier, Python, the Claude API, Airtable, Notion, custom Next.js apps. Whatever actually fits the problem. I'll tell you up-front when no-code is the wrong call for the job.

The product is rarely a single script. It's usually a few chained scenarios with the right human approval gates in the right places — automation where it's safe, human judgment where it matters.

See the cases for what this looks like in practice.

/ principles I keep coming back to
  • Mechanism > restriction. When a workflow misbehaves, the durable fix is a structural guard — server-side validation, a tool schema, a confirmation gate — not a polite rule that hopes the AI or the user cooperates.
  • Human approval where it matters. The valuable judgment calls stay with the human. Everything upstream and downstream of those calls is automatic.
  • Idempotent by design. Workflows you can safely re-run mid-day are workflows you can actually debug. Everything else turns into folklore.
  • Honest about what NOT to automate. Some processes are doing real work in their slow, manual form. I'll tell you when that's the case instead of selling you a build that hurts.
/ how to engage me

Two ways to work together. Both quoted up-front after the diagnostic — no surprises:

option 01 · ai coaching

AI coaching for SMB owners

Specifically for small business owners who feel lost about AI. 1:1 sessions in plain language — English or 中文 — where we sort out what AI can actually do for your specific business, what's hype, what isn't, and where to start. No tech background needed, no jargon, no tool tutorial — this is orientation, not training. You walk away with clarity instead of more noise. 45-min pre-paid sessions.

option 02 · case build

I build the whole system

End-to-end build, tested, documented, handed over with full source. Fixed build fee scoped per project. Optional monthly hosting / maintenance retainer if the system needs ongoing operations (most don't — once it's running, it runs). 1-2 week typical delivery, 3-month bugfix window included. Best when the workflow is load-bearing and you don't have the bandwidth to own the build.

Not sure which one fits? The free diagnostic ends with that question for each opportunity — DIY or hire — so you don't walk into the conversation guessing.

/ get in touch

Fastest path: run the diagnostic. I'll see your responses on the other side and reply within 24 hours with a specific direction.

Or email me directly at hello@michaelzconsulting.com. Either way, you talk to me — not an intake form, not a sales rep.

— Michael Z · Vancouver · workflow automation consultant