Anthony Lee
Lab · Working notes

Experiments in operations, technology, and AI.

A working record of what I'm building, learning, and applying. Some become production tools; all sharpen how I solve problems.

// intro
A short note on how this works

I learn fastest by doing. Whenever I run into a repetitive task, an awkward workflow, or an interesting new technology, my first instinct is to experiment with it.

Sometimes those experiments grow into production systems. Other times they just help me understand what's possible a little better. Either outcome is useful.

This page tracks those explorations — the same areas I bring to an operations or technology role.

Experiments

Selected experiments.

Problem

A personal site is small on paper, but the details compound — design system, routing, animations, SEO, metadata, MCP integration. Historically that's a weekend of setup before you get to the actual craft.

Approach

Used Lovable as the development environment and AI as a collaborator: sketching structure, iterating on copy, refining components, and reviewing every decision. The AI accelerated implementation; direction, taste, and editing stayed with me.

Outcome

A site I own end-to-end, built in a fraction of the time it would have taken by hand, with a workflow that lets me keep improving it.

Lessons learned
  • AI removes the boring parts of building, not the thinking parts.
  • The quality of the output tracks the quality of the questions you ask.
  • Small, considered iterations beat one big prompt every time.
Problem

The existing site was slow to update, tied to outside developers, and drifting away from the brand. Every small change became a project.

Approach

Rebuilt from the ground up with AI-assisted development, keeping the stack simple enough that non-specialists could reason about it. Prioritized performance, clarity, and a workflow the team could maintain in-house.

Outcome

A modern site that ships changes in minutes, no longer depends on vendors for everyday updates, and looks like the brand it represents.

Lessons learned
  • Ownership is a feature. In-house control beats a fancier tool you can't touch.
  • The right stack is the one your team can maintain on a Tuesday afternoon.
  • AI made a project this size feasible for a team our size — that's the real unlock.
Problem

The existing setup lived on top of third-party tools. It worked, but the seams showed at every step of the customer journey, and every improvement was gated by someone else's roadmap.

Approach

Rebuilt end-to-end with a focus on the customer's actual path — browsing menus, managing an order, getting support — while keeping operations simple enough that a small team could run it without heroics.

Outcome

A smoother checkout, fewer support handoffs, and a platform the business could iterate on its own timeline.

Lessons learned
  • Third-party platforms are a great starting point and a poor endgame.
  • Customer experience is mostly the invisible work between the visible steps.
  • Simplicity for the operator quietly becomes simplicity for the customer.
Problem

AI coding tools reward the people who learn to use them well and quietly punish the ones who don't. The gap keeps widening.

Approach

Use AI across the full loop — prototyping, writing, reviewing, debugging, refactoring, learning unfamiliar tech. Stay close to the underlying problem so the model has a useful partner, not a blank check.

Outcome

A working practice, not a finished one. Faster iteration, less time stuck, and more headroom for the harder decisions.

Lessons learned
  • AI is a collaborator. Understanding the problem is still your job.
  • The best prompt is usually a well-scoped problem, not clever wording.
  • Read every diff. The muscle you don't use is the one that atrophies.
Problem

Most business dashboards answer questions nobody asked. The interesting numbers get buried, and the ones on screen quietly stop being trusted.

Approach

Start from the decisions leadership is trying to make, work backwards to the questions, and only then design the chart. Cut anything that doesn't move a conversation forward.

Outcome

A slower, more deliberate design process — and dashboards people actually open on a Monday.

Lessons learned
  • Understanding beats reporting.
  • Fewer, sharper numbers beat more numbers.
  • If nobody argues about a chart, nobody is using it either.
Problem

Most repetitive work isn't dramatic enough to make a project out of, so it just keeps happening. It adds up faster than anyone admits.

Approach

Watch for the tasks that keep coming back, then remove them one at a time — a scheduled report here, a notification there, a cleanup script that runs while nobody's watching.

Outcome

A slowly growing catalogue of small wins. Nothing worth a press release; a lot worth an afternoon.

Lessons learned
  • Boring automations pay back the most.
  • If it's worth doing twice, it's worth looking at.
  • Automate the task, protect the person.
Notes

Working notes.

Short notes on operations, technology, and AI.

Note 01Draft

AI is an operational multiplier for small teams

A two-person team with good AI tooling can now do work that used to require five. The leverage isn't in replacing people — it's in removing the bottlenecks that slow them down.

Note 02Draft

The best tools are invisible

People don't love software because it has more features. They love it because it gets out of the way. Removing friction is harder than adding functionality — and almost always more valuable.

Note 03Draft

Every business is becoming a systems business

Whatever you sell, your competitive edge increasingly depends on how well your internal systems run. Operations, technology, and customer experience are no longer separate departments.

Note 04Draft

Ship the rough draft, then iterate

You learn more from a working version in the world than a polished one in your head. Start small, use it, then decide what actually needs to be good.

Now

Current focus.

Where my attention is right now. These shift as projects and interests evolve.

  • AI-assisted development workflows78%
  • MCP servers55%
  • Agentic AI40%
  • Modern React architecture70%
  • Automation65%
  • Analytics60%
  • Prompt engineering72%
  • Product thinking68%
Roadmap

What's next.

Areas I want to explore and problems I want to solve.

01Restaurant inventory forecasting
02AI-powered operations dashboards
03Internal knowledge assistants
04Scheduling optimization
05Business analytics experiments
06Customer journey mapping
07POS integrations
08Automation templates
Closing thought
"The best systems are the ones people stop noticing."