Systems

Websites, workflows and intelligent tools built around how information moves. I structure information, journeys and workflows, and turn them into digital experiences people can use.

Discuss a project

What this means in plain language

Deciding what to say, to whom, in what order and in what format, so the content is doing a job rather than filling a feed.

Content strategy

Deciding what to say, to whom, in what order and in what format, so the content is doing a job rather than filling a feed.

Systems thinking

Looking at how the parts of something affect each other, so a change in one place is made with the whole in mind.

Featured work

All work in this line

Explore the workflow concept and the self-initiated browser instruments.

Capabilities

01

Websites & interfaces

Structure, navigation and digital experiences designed around how people actually use information.

02

Workflow automation

Connecting repetitive steps and tools so routine work can happen automatically.

03

Data collection & distillation

Gathering fragmented information, cleaning it, and reducing it to useful signals.

04

AI knowledge systems

Connecting AI to relevant information, so an answer is grounded in a knowledge base rather than improvised.

05

Agent workflows

Multi-step AI-assisted processes where tools, rules and human review each do their part.

06

Custom AI skills

Reusable task-specific capabilities, instead of writing the same instructions again each time.

07

Custom integrations

Scripts, APIs and lightweight logic that connect systems when off-the-shelf tools cannot.

Tools & implementationPython · n8n · Dify · RAG · APIs · scripting · scheduled jobs · data pipelines

Current R&D

Ongoing experiments in automation, conversational AI, and systems that turn fragmented information into useful actions.

01

Cantonese conversational AI

Exploring how customer-service systems can better understand Cantonese intent, context and natural conversation.

02

AI automation

Testing how agents, scheduled processes and connected tools can move information and actions through a workflow.

03

Data intelligence

Exploring collection, filtering and distillation methods that reduce large information sets into useful signals.

04

Custom agent skills

Experimenting with reusable skills, tools and instructions that let AI agents perform specialised tasks more consistently.

What gets delivered

01

Map

A clear map of who does what and when.

02

Routine

A working draft-to-release routine.

03

Walkthrough

A walkthrough for the people using it.

How we get there

01

Agree the scope. Decide what the first piece needs to do.

02

Make and review. Keep one clear approval point.

03

Hand over. Leave the next step clear.

How I think

01

Observe

Watch the work as it happens, who does what, when it ships, where it stops.

02

Diagnose

Separate a process problem from a decision problem. Most stalls are decision problems.

03

Structure

Name three moments: what goes in, who approves it, and when it ships.

04

Build

Build on what the team already uses, rather than a new stack.

05

Iterate

Run one cycle, see which step nobody follows, then tighten that one.

Do I need to commit to a big project?

No. We can scope the smallest useful piece first. You see the deliverables and price before agreeing to paid work.

What if the direction is not right?

We agree a review point and what can change before production starts. That boundary goes into the scope.

We already have someone doing this.

I can take on a defined part alongside your team. We agree who owns each decision and handover.

How we could work together

Send me one brief. I'll turn it into a clear scope, deliverables and price before you commit.

  • 01Brief
  • 02Scope
  • 03Agreement
  • 04Start

Paid work begins only after the scope and fee are agreed.