Digital tools, built and handed over
When advice needs a tool, I build it: a small app, a dashboard or an automation, with plain instructions so your team owns it.
Built so far: a health directory, a cycling app and a pet-travel map.
Clear decisions, steady growth.
Reed Boat is an independent advisory practice for small and growing businesses. I help you decide what to do next, then set up the numbers, processes and tools to make it stick.
The first call is free, with no obligation.
How I can help
Four kinds of work. Start with one, and add another when it's useful.
Operations and workflow
Map how the work really runs, where the hours go and what gets done twice, then choose what to fix first and what to leave alone.
Tool: Workflow Map, a short survey your team fills in that shows where the week goes.
Data and reporting
Turn the numbers you already keep into decisions. Clean records, a few measures that matter and a simple monthly view.
Recent: public registers turned into one clean, searchable dataset.
Strategy and go-to-market
Decide who you're for, what to offer first and how to reach them, with a launch plan your team can follow.
Recent: who to sell to first, and the launch plan, for an early-stage startup.
Selected work
Products and tools I designed and built. Each one answers a question people couldn't answer clearly.
Iasis · Live
Is this practitioner really qualified?
7,477
practitioner records checked against professional registers
Illustration, not real data
Read the case
- Before
- Patients had no easy way to tell who was registered.
- After
- Each practitioner checked against professional registers, in one searchable directory.
- Problem
- Patients can’t easily check whether a traditional medicine practitioner is registered with a recognised professional body.
- Approach
- Pull registration data from the professional registers and republish it as one searchable directory, then make those pages easy to find in search and AI answers.
- Status
- Live and in active development.
Workflow Map · Available as a service
How does this business really run?
1 week
from team survey to a ranked list of what to fix first
Illustration, not real data
Read the case
- Before
- Nobody could say where the week's hours went.
- After
- A clear list of what to automate and what to keep by hand.
- Problem
- In many small businesses the critical work lives in one person’s head, undocumented, and stops if they leave.
- Approach
- A short survey the team fills in maps how the week really runs, where the risks and bottlenecks are, and what to automate first.
- Status
- Available as a service.
- Earlier version
- I first built this as a software product called AkiraOS, which I no longer sell. The original site is still online as it was, including its old name, pricing and examples. See the original AkiraOS site
Ding Ding YYC · iPhone app in progress
Which route can I bike in the winter?
4
city data layers in one map
Illustration, not real data
Read the case
- Before
- Closures, comfort and snow clearing sat in separate datasets.
- After
- One map that says whether a route is rideable now.
- Problem
- Cyclists can’t easily tell whether a route is comfortable, closed or cleared of snow today.
- Approach
- Combine the city’s open data on comfort, closures, snow clearing and lighting into one map, with each layer’s source and age shown.
- Status
- iPhone app in progress.
PetPawsible · Finished
Will this place let my dog in?
1,000
places checked in 3 countries, with suggested walking routes
Illustration, not real data
Read the case
- Before
- Pet rules were scattered across reviews and websites.
- After
- One map of pet-friendly places and routes.
- Problem
- Travelling with a pet means guessing which cafés and restaurants will let your dog in, especially in a new city.
- Approach
- A map-first app of pet-friendly places and walking routes, starting in Europe.
- Status
- Finished and no longer updated.
Have a question like these? Book a 30-minute call
How I work
Short steps with a fixed scope, agreed before each one starts.
Week 1
Understand
Conversations with you and your team, and a look at the numbers you already have.
Week 2
Decide
A short written plan: what to do first, what it costs and what to leave alone.
Weeks 3 to 6
Set it up
The process, report or tool that makes the decision stick, built with your team.
A month later
Review
A check on what changed, what still rubs and what's worth doing next.
Thinking
Short notes on the questions behind the work. Open one to read it here.
The workflow I use to build faster with AI design toolsEach tool is a specialist, not a one-shot answer, and a real design system is what makes it work.
Every “AI can build your app now” pitch skips the part that matters: knowing which tool to use at which stage, and why. I treat each tool as a specialist rather than a one-shot solution, and it gets me from a vague idea to a production-ready system faster and cheaper than any single tool end to end.
Ideate. Rough the idea out loose and cheap; the output is a spec to react to, not a deliverable. Borrow a system. Pull a real design.md from someone else’s actual design system, not vibes. Build for real. Hand both to Claude: what to build, and how it should look. Bring it home. Move the real build into your own repo, conventions and infrastructure.
The design.md handoff is the actual unlock. Most AI-built sites look AI-built because nobody gave the model an opinion about design. Give it something to follow, a real constraint, an opinion, and the default disappears.
Borrowed intelligenceUsing AI to build is a loan, not a gift, and loans get paid back.
Using AI in building a product is like borrowing knowledge from an external partner. Because it’s a borrowing, you’ll need to pay it back someday, and that matters when you plan for cash flow and for scaling up.
It doesn’t feel like debt in the moment. A model drafts your onboarding flow, your database schema or an agent pipeline in an afternoon that would have taken a week. That’s real leverage, but leverage has to be serviced.
The bill comes due in the maintenance of code you didn’t fully design, in debugging a system that made decisions faster than you could review them, and in paying for that capability every month instead of hiring it once. None of that means don’t borrow. It means plan the repayment before you take the loan, not after it’s due.
The underrated power of boring data infrastructureThe unglamorous plumbing is usually what decides whether a product works.
The parts of a system that get the least attention, schemas, pipelines, validation and naming things consistently, are usually the parts that decide whether everything built on top of them actually works.
Boring infrastructure isn’t a placeholder for the interesting work; it’s the thing that makes the interesting work possible.
Why a reed boat
A reed boat is thousands of thin reeds tied into one strong hull. Good advice works the same way: many small, clear decisions, tied together, that carry a business forward.
I come from finance and data, so I start from the numbers and keep the advice practical.
Thu Trinh, founder of Reed Boat. LinkedIn