Sixty Seconds, Second Place, and the Ten-Year-Old Idea Behind chequeup

I did not go there to pitch anything

On a sunny Saturday afternoon I walked into NeighbourSpace on Douglas Street for the Victoria edition of Shipaton, RevenueCat's global hackathon for shipping a brand-new app to the app stores. ^[1]

I went for the reasons the listing advertised. Meet some Island developers, hear how Shipaton works from Aleksandra Komagorkina(opens in new tab), eat the free snacks. No finished idea, no team, no experience required.

About thirty-five of us turned up. Partway through the afternoon the floor opened. Sixty seconds each, slides and a demo optional, and three questions: why do you want to build it, who is it for, what will it do. A few people had something running. Most had an idea and nothing else. I was in the middle, with no pitch prepared but a partially built website.

Thirteen pitches went up, from fifteen people. I was not on that list when it started, but I put my hand up late and went second to last. I pitched chequeup, and the pitch was that your take-home pay could be bigger. Sixty seconds later I sat down to a room of blank stares, which was entirely fair. T4 slips, RRSP contributions and reducing deductions at source sit right on the border of making a person's head explode.

There were no judges and no panel. Everyone voted, in a live poll on the screen at the front, for the app they most wanted to see shipped. Himaja Mankala(opens in new tab) took first with Pitch, Please!, a genuinely brilliant idea I cannot wait to see in production. I came a close second, which the blank stares had not prepared me for. We both went home with the same trophy.

The idea I have been carrying since 2016

The pitch, more or less as it came out: Canadian T4 employees can legally keep more of every single paycheque by having less tax taken off it.

That is not a loophole, and it is nothing aggressive. It is timing.

Your employer calculates the income tax to withhold from each pay and remits it for you, along with Canada Pension Plan and Employment Insurance. ^[2] That calculation assumes a plain financial life. It does not know about the RRSP contributions you make on your own, your child care costs, or other credits you are entitled to, so it withholds as though none of them exist, all year.

The Canada Revenue Agency is not hiding the alternative. It publishes Form T1213 for people seeking "reduced tax deductions at source for any deductions or non-refundable tax credits that are not part of the Form TD1". ^[3] You apply, the agency issues a letter of authority if approved, and your employer reduces the tax taken off each pay. The tax you owe for the year does not change. What changes is whether that money sits in your account or Ottawa's while you earn it.

This is not obscure. Of the returns assessed between 7 February and 4 August 2026, 18,835,430 came back as refunds, worth $45,027,914,000. ^[4] Over-withholding is only one of several reasons for a refund, so not all of that should have stayed in people's pockets. But a refund is not a windfall. It is your own money, returned months later and in the ordinary course without interest.

I worked this out for myself more than twenty years ago. Ten years ago I worked out the second half: a model designed so the money the first part frees up does more than simply arrive earlier. That is why chequeup is a product rather than a blog post, and it stays mine for now.

None of this is tax, financial or investment advice, and chequeup will not be giving any. Everyone's circumstances differ, approval is not automatic, and the agency says it will not usually issue a letter of authority to someone who has an amount owing or returns outstanding. ^[3]

Why a good idea sits in a drawer for ten years

It was never the idea. I have known exactly what to build for a decade.

It was the build. Two app stores and two review processes. Tax logic that has to be right every time, because being approximately right about somebody's pay is worse than useless. An interface people will trust with their income. Onboarding, support, privacy, payments.

None of that is mysterious to me. I have built versions of all of it, which is precisely the problem: I knew exactly how much work it was, a remarkably effective way to never begin. Every year or so I would price it in evenings and weekends, look at the number, and put it back. The idea was good. The economics of one person building it were not.

What actually changed

AI-assisted development changed them, and I want to be specific, because this claim is usually made badly.

It does not make me a better engineer. It collapses the cost of the work that never needed my judgement. The blank file. The scaffolding. The boilerplate. The part of any project outside what I already know well, where I used to lose a weekend learning something I would use exactly once.

The gap it closes is not between not knowing and knowing. It is between knowing and having the time.

What it does not do is decide what to build. It will not tell you the answer it just produced is subtly wrong, or notice that the feature you asked for should not exist. It has no taste. It will build whatever you describe, cheerfully and quickly, including the wrong thing, beautifully.

Thirty years is the unfair part, and I should say so

For a couple of years I have been saying in public that AI will open production up to people locked out of it: those without the physical capability, the formal training, or twenty years of scar tissue.

But I am not the underdog in that story, and it would be dishonest to write this post as though I were. I have been in technology for more than thirty years, and building for the web since before this century started, more than a decade before Chrome existed and nearly two decades before Edge launched. When I use these tools, they multiply something that took decades to accumulate.

Sean Charman is the better example. He roofs houses here in Victoria, and recently launched Hobby Card Index(opens in new tab), which indexes millions of trading cards and prices them off real sold data. Not a landing page. A working product, built by someone whose day job is on a ladder.

That is the part the enthusiasm skips, in both directions. The tools multiply what you bring. Sean brought a collector's obsession and knew exactly what was missing. Bring none of that and they will only help you build the wrong thing faster.

I got my own preview earlier in the year. Two of us built Queue Smart, an emergency room patient communication platform, in six hours at a hackathon, and it won. That one ran on low-code tooling, not AI. Different leverage, same arithmetic, and the moment the numbers on chequeup started to look different.

What sixty seconds taught me

I had never once said this idea out loud to strangers. Sixty seconds taught me more than the decade did.

Sixty seconds is the best editor I have ever had. Three questions, sixty seconds, no slides to hide behind and no roadmap to gesture at. Everything that was not load-bearing fell off on the walk to the front of the room.

The sentence that worked was the simplest one available. Not the model. Not the algorithm. What landed in that room was that people like paying less tax and having more in their digital pocket. I had spent a decade polishing the wrong sentence.

The hard part of chequeup is not the engineering. I went in braced to defend the build, and realised the questions worth answering belong to the person using it, and not one is technical. Is this allowed? Will their employer think it odd? What happens if they get it wrong? That reframing is worth more than the trophy.

Going late is an advantage, and second place is data. I heard eleven pitches first, and could feel what the room responded to. Then about thirty-five people with no context voted it up on sixty seconds and no visuals. The cheapest validation I have ever bought, and I did not go there to buy it.

The readiness I had been waiting for was a decision. I decided to pitch roughly four minutes before I pitched. The condition, it turns out, was putting my hand up.

22 August

On Saturday 22 August I will be at Cursor Codechella here in Victoria, a free one-day hackathon that ends in a hard submission deadline, three-minute demos and judging. ^[5] One prize track runs through to Shipaton, which accepts apps first released between 1 August and 30 September and advertises over US$700,000 in cash prizes. ^[1]

The plan is to walk in with the idea and walk out with chequeup built and submitted. The app stores take their own time, and the Shipaton window runs to the end of September, so that is the real deadline. Ten years of not building it, and now a Saturday.

That is rather the point. The reason it took ten years was never the idea and never the knowledge. It was the cost of the build, and that cost has fallen through the floor.

Wish me luck.

If you have one of these in your own drawer

Most people I talk to have one. A thing they have thought about for years, know is good, and have never built, because the distance between the idea and a working product was too far to cross in evenings.

That distance is much shorter than it was, and it is shorter for your business than it is for me on a weekend project. This is moving faster than most people realise: 19.2% of Canadian businesses reported using AI to produce goods or deliver services in the twelve months before being surveyed in the second quarter of 2026, three times the share two years earlier. ^[6] The same shift that makes chequeup buildable makes the internal tool you keep postponing, or the customer-facing thing you have described to your team a dozen times, genuinely achievable now.

Taking something a person has been carrying for years and getting it built is the part of this work I like most. Sometimes that means building it. More often it means teaching a team to use these tools with enough judgement that they can build it themselves, which is the more valuable of the two. Both are what Grey Sky Tech does for clients, through artificial intelligence advisory, training and custom builds.

So if you have one in the drawer, get in touch and tell me what it is. Worst case, you get sixty seconds of my full attention, which I can now confirm is quite a lot.