unofax.com

I've been writing software for 25 years, and been getting paid for the last 20. My AWS account will be turning 18 this September, and it should be quite the celebration. If I may toot my own horn: I've got pretty good at writing and deploying software in that time - to the point that you're actually spending your time reading this blog. All this, and the release of Opus 4.5 in November 2025 hit me like a tonne of bricks. My value-add (as I knew it) would soon be over.

Is that a dramatic enough lede? I've also been blogging about software for at least 17 years and I still feel like I haven't got the hang of it. In any case, Opus 4.5 was huge. Let me set the scene: after my partner had been using it for a few months, I had finally got around to installing Claude Code in mid-November. Opus 4.0 seemed too slow and expensive to use, and Sonnet 4.5 was the default model. I gave it a description of a project I wanted to build: a combination of standard AWS crud stuff and some more arcane network packet-wrangling and little-used CLI plugin standards. It generated some surprisingly good code. Not perfect, had a show-stopping (but fixable) bug, but definitely usable. I thought this could save me some time.

A week later Opus 4.5 is released, becomes the default model and changes everything. It fixed all the bugs I identified in the code written by Sonnet, plus a handful of bugs I hadn't yet even noticed. It was objectively a better programmer than I am after two decades of being hands on the tools. I've always been in agreement with Ben Kehoe ever since I heard him state (paraphrasing) that your job as a developer isn't to write code, it's to solve problems. As much as I didn't want this to be the case, it's true. The thing that had always most distinguished me from my peers (memorising an unreasonably large surface area of AWS APIs, being able to write ~complete serverless apps and infra without consulting docs, etc) is now a prompt away.

Anyway, that's a very long way of saying: AI is here, it's very good at generating whatever code you ask it to, and it's only ever going to get better. So I need to start finding a new way to be useful. My partner and I decided that we should take one of my many SaaS ideas from over the years and actually try to ship it. It's a pretty big idea, and I figured we'd probably fail, so instead I suggested that she start with some small ideas, complete them, move onto bigger ideas, learn and iterate in a snowball fashion. She had just quit her full-time job, so it was perfect timing.

And that's what we did: she made some very impressive small projects, shipped them to production and learned a lot about how to work with AI effectively. Then we decided our next project should be a complete SaaS: users can land on a marketing site, use the functionality and pay for it. And that we'd build it entirely using AI.

So that's how we made unofax.com. Payments are received using Square (my day job), hosted on AWS Lambda and AWS Step Functions (my favourite tech stack) and the frontend is Vite? We outsource the actual fax-sending to another company that does a great job, but requires maintaining a balance to send faxes. So our unique differentiator is that we don't require any sign up and you can pay per page. Somehow, we've already made a few sales, adding up to a few dollars. That's the most surprising part to me.

Will it allow us to retire? Very unlikely. It serves more as an excuse to learn how to ship something end-to-end, and to learn SEO, marketing, frontend dev, customer support, etc. It also taught us what pain points we should expect when using AI, and provides an ideal proving ground for trying new ideas to avoid them. Because it costs us close to nothing to run, we can also keep it running indefinitely. It has also provided an excellent and humbling opportunity to learn quite how inelegantly I can express my ideas, compared to where I want to be. So I can work on that too.

What's next? I realised while writing my last blog post that it'll still take a while before AI really understands AWS security in depth: it kinda sucks at SCPs, RCPs and nuanced IAM. So I've probably got a day job for a while longer. We're also starting work on our next idea, slightly bigger, more ambitious and more relevant than a fax-sending service in 2026. Maybe that idea will earn enough to pay for lunch every day. And then I can move onto the idea after that.