Transmitting from 47.5615° N, 52.7126° W

We build the AI system your organization actually needs.

Not advice about AI. Not a tool licence. We build — a public research archive people can question in plain language, dispatch software for crews in the field, back-office tools for a brokerage. Different sectors, same method: understand the work first, then build for it.

Based in St. John's — building for organizations across Canada

Fall 2026

Selected for Movement51's Founder Lab — one of 20 founders nationally, and the only company from Newfoundland and Labrador in the cohort.

Where we come in

Tell us what it should do. We'll build it.

AI is useful to almost every kind of organization now — a government department, a fishing enterprise, a clinic, a construction firm, a research group, a museum. The limit isn't your sector or your size. It's that most people can picture the outcome they want without knowing what would have to be built to get there.

That gap is our job. You describe what should happen — the report that writes itself, the archive anyone can question, the schedule that stops falling apart, the thing you've assumed was out of reach. We work out what it takes and we build it.

You picture it

“I wish we could just ask our own records a question.”

We work it out

What data exists, what's realistic, what it costs, and whether it's worth doing at all.

We build it

A working system your team uses, not a report about what you could theoretically do.

The raw material is usually already there — maintenance logs and inspection reports, sensor readings and dispatch records, case files and permit histories, technical drawings, survey data, call recordings, decades of archived documents, and yes, the spreadsheets nobody has time to open.

We work with businesses, government departments, agencies, and non-profits. The method doesn't change much between them. The work does.

Government and public sector Business and industry Healthcare and clinics Trades and field services Energy and resources Marine and fisheries Research and academia Tourism and hospitality Professional services Real estate Non-profits and associations

Five ways that usually looks in practice

Automate

Find the repetitive work and take it off people's desks, so their attention goes where it's actually needed.

Connect

Get systems, information and applications talking to each other instead of around each other.

Understand

Make the information you already hold easy to find, ask questions of, and act on.

Build

Create custom tools, assistants and applications when nothing off the shelf is good enough.

Improve

Look at existing software and processes with fresh eyes and ask where new technology makes them better.

The technology is only useful if the result is useful.

AI should help your people, not replace them

The goal is taking away the work nobody wanted to do, so your team can get on with the work you hired them for.

Fix the process, don't decorate it

Putting a chatbot on top of a broken process just gives you a faster broken process. We'd rather sort out the process first.

How it works

Three steps, and you can stop after the first.

01

We take a look

We go through your existing data and workflows and tell you where AI would genuinely help — and where it wouldn't. Free, no obligation, no pitch at the end unless there's something worth pitching.

02

We try one thing

One working system against one real bottleneck. Small scope, low risk, and you can see for yourself whether it made a difference before committing to anything larger.

03

We build out what worked

Expand it, connect it to the tools you already use, and make sure your team can run it without us standing over their shoulder.

Featured work — delivered

Churchill Falls Information Assistant

Decades of work on the Churchill Falls agreement — economic assessments, legal documents, historical records — existed, but there was no practical way in. A journalist on deadline or a student writing a paper faced a wall of PDFs.

We built a bilingual research assistant across the full body of material — long-form energy contracts, technical assessments, decades of economic analysis and the historical record around them. Three ways to ask: Voice answers in Dr. May's own voice, Fast returns a focused text answer, and Deep runs comprehensive research across every source, with citations you can open.

Work that was locked in documents is now something anyone can just ask about.

Visit churchillfalls.info →
The Churchill Falls Information Assistant landing screen, offering three ways to ask a question: Voice, which answers in Dr. Doug May's own voice; Fast, for focused text answers; and Deep, for comprehensive research across all sources.
Three ways in: hear it in Dr. May's voice, get a fast answer, or search the full library.

A pattern we build well

We turn archives into things you can talk to.

Almost every long-lived organization is sitting on one: decades of documents, recordings, case files or correspondence that hold real answers and can't practically be searched. A keyword box finds a filename. It can't tell you how a position shifted over twenty years, or that a note from 2003 bears on something said last week.

We build research companions that can. They read the whole collection, answer in plain language — in the author's own voice, if that's useful — and cite sources so anyone can check the work and go deeper.

Delivered

Contracts and analysis, two languages

Dense energy agreements, technical assessments and decades of economic analysis of Churchill Falls, built for Dr. Doug May — now answerable in English or French, by voice or text.

In development

28 years, text and audio

Unknowncountry's full archive since 1998, including decades of recorded interviews — searchable as a single body of work for the first time.

The same build fits university special collections, museums, records offices, professional associations, law firms, and any organization whose institutional memory currently lives in one or two people's heads.

Also underway

Three builds in progress.

Some work starts with a client's problem. The rest starts with us noticing something nobody has fixed yet — a way of working we've watched up close, and a fair idea of what technology could do about it. We build those anyway, without waiting to be asked. It's how we find out what's actually possible, and it means the next organization that has the problem doesn't have to start from nothing.

Client build · in development

Unknowncountry Research Assistant

Whitley Strieber's Unknowncountry has published continuously since 1998 — thousands of articles, journal entries and firsthand accounts, alongside decades of recorded interviews. It's one of the largest archives of its kind anywhere, and almost none of it is findable. We're building a research companion that answers in Strieber's own voice with sources cited, and can follow a single thread across the whole archive — connecting a journal entry from 2003 to something recorded last week.

Our own product · waitlist open

Agents4Agents

An assistant that answers an agent's phone at eight on a Saturday night. It connects straight to listing data, talks to buyers by voice or text in any language, answers the neighbourhood questions people actually ask, qualifies leads and books showings — then emails the agent everything. No client asked us for this one. We saw the opening and built it, and it's taking founding members now.

Our own concept · in development

Traverseclean

Scheduling and tracking for field service crews — the recurring visits, the shifting sites, the paperwork that proves the work got done. Nobody commissioned this either. We'd seen how the day actually runs for teams working across multiple locations, and how much of it still depends on someone holding it all in their head.

A healthcare build begins this fall. More on that when we can say more.

About — Lillian Fidler, callsign VO1XYL

Newfoundland has been early to new technology before.

Lillian Fidler on Signal Hill in St. John's, speaking into a handheld radio microphone on a fiercely windy March day, hair and hat caught by the wind, with Cabot Tower behind her.
Lillian Fidler (VO1XYL) on Signal Hill, March 2024 — a bitterly windy day, transmitting to the International Space Station from the headland where Marconi received the first transatlantic signal in 1901.
MindStudio Level 3 HarvardX CS50 AI 25+ years building Amateur radio — callsign VO1XYL
1901

The first signal

Marconi received the first transatlantic wireless signal on Signal Hill, a mile from where we work.

The early days

A programming student, and one stubborn idea

Studying computer programming at the University of Alberta, Lillian took on a project examining a real business to work out where automation could help and what it would be worth. The habit stuck: don't start with the technology, start with the problem.

2000s

First to merge design and computing here

Lillian Fidler built one of St. John's first studios to bring computer technology into graphic design. Six nominations for Graphic Designer of the Year followed, for album cover work.

2000s

Newfoundland's first podcast

Co-created Republic of Avalon Radio, before most people had a word for what it was.

2024

Talking to the space station

Coordinated a live radio contact between Grade 9 students in St. John's and astronaut Loral O'Hara aboard the International Space Station. Four schools took part, tied into the Grade 9 science curriculum. Read the coverage →

Now

The same instinct, applied to AI

Building working systems rather than slide decks — including the Churchill Falls Information Assistant, built for economist Dr. Doug May.

Get in touch

Start with a conversation, not a contract.

Maybe a process takes too long. A system doesn't talk to another system. Information is buried. Or you've been working around the same problem for years. You don't need an AI strategy and you don't need to know what technology you need — bring us the problem and we'll start there.

The AI Opportunity Assessment is free. Some organizations walk away with a project. Some walk away knowing they don't need one yet. Both are fine by us.