memorydial

Last Sunday, I resurrected my dead startup in four hours. MemoryDial was a web monitoring service I built in 2014, and abandoned in 2016. It tracked changes on any website, any element, and sent you notifications when something updated. Think of it as a universal notification system for the web, back before push notifications existed. Claude Code got it running again before lunch. What took me six months of nights and weekends in 2011 took a single morning in 2025.

I wasn't planning to resurrect it. I have been using Distill.io every day for web monitoring for the past few years, circa 2018, their browser extension is really good. It has a CSS picker, and it watches websites for changes on a cadence you select and it alerts you when something updates. It's exactly what I built in 2011. But now I need something I could hook into my AI systems with no API limits and no constraints on what I monitor or how. That was when I realised that MemoryDial was sitting there dead in a Git repo.

I fired up Claude Code, ran /init and four hours later, it was running locally again. The Django version was 1.3! Claude updated it to modern Python. It didn't rewrite anything, it just updated deprecated dependencies, swapped out old libraries. The core logic still worked. The idea still worked.

But this isn't really about resurrecting old code. It's about a pattern I kept missing for fifteen years.

The Milkman at 3am

Let me tell you about my actual first attempt at building something. Before MemoryDial.

I was coding late one night when I heard the milkman arrive. Our house was on his route. Ma would always scramble for cash when he knocked, digging through drawers for exact change. I'd just built an online ordering system for a college assessment. I was exploring payment processors in that pre-Stripe world of AIB Merchant Services and Realex Payments. I thought I could help him out.

When he came to collect money one evening, I asked about his route. “Want to come along?” he said.

At 3am, delivering milk in south Dublin, I learned what the job actually involved. Spider webs stretched between every fence and bush, exactly at face height. “Ah fecking spiders are everywhere tonight,” he said, wiping web remnants from his face for the twentieth time that hour. But the spider webs weren't the real problem.

He'd been robbed three times collecting money. Once, he caught the guy. His son helped him find another. The third got away clean. His whole sleep pattern was destroyed. Not by the early deliveries but by the afternoon money collection. Sleep a few hours after morning deliveries. Do commercial runs. Then try to sleep again after chasing payments. Biphasic sleep ruined by people saying “get me next week.”

This was 2009. I saw the opportunity immediately. Online payments would solve everything. No more cash collection, no more robbery risk, better sleep. He would have easily paid more than twenty euro a month just for that.

So I built MilkDirect. Customer portal with order management. Milkman dashboard for route planning. Admin panel for the dairy. Holiday pauses. Calendar views. Special items. The works.

He needed only required payment processing. I built an entire dairy management platform, kitchen sink included.

I kept thinking it wasn't complete enough. It wasn't perfect enough. I wanted to go back to him with the full system, not just the payments part he actually needed. By the time I got it finished in early 2013 I were too late. MyMilkman.ie launched in the summer of 2013 by Glanbia. I had our version with payments working in 2010, but faffed about trying to deliver the perfect product for another 4 years. Four years. I had a them four years before.

Chrome's Dead Tiles

By 2014, I'd moved on to MemoryDial. The idea came from staring at Chrome's “Most Visited” tiles every day. Eight static squares showing dead screenshots of websites. They should be alive. They should pulse with changes like Facebook's red notification bubble. That addictive dopamine hit, but for any website.

This was my pitch from 2014:

“Memorydial is a service that allows users to get dynamic notifications from websites. Say for instance you want to know when a particular website you frequent gets updated. How do you currently do this? Do you get bombarded with RSS updates? Do you wait for a friend to share it with you?”

I thought I were solving how people discovered content. “Users can define websites or portions of web pages that they wish to be notified about.” The technical implementation was solid. WebSockets for real time updates. Diff algorithms to detect changes. Django handling the backend. You could track any element on any page. I even had grand visions. Group notifications because “we believe this to be a compelling service to users based on the feedback we have received.”

What feedback? We'd talked to maybe three people. Friends who said “yeah, that sounds cool bro.”

I built the whole thing first. Then tried to figure out who needed it. I pitched to NDRC accelerator in Dublin, down at St James's Gate. Met with advisors. I did get some solid advice, “Focus on ideas where you already figured out where the customers are...” I'd built it assuming they would come. Field of Dreams thinking. Kevin Costner in a cornfield, except it was Django and no jazz in Dublin.

The Pattern I Couldn't See

Both times I could spot the opportunity. The milkman's spider webs were a real problem with a real solution. Chrome's static web tiles were begging to be dynamic.

Both times I could build the technical solution. MilkDirect worked. MemoryDial worked. Both times I failed for the exact same reason.

When building takes six months of nights and weekends, you have to commit before validating. You invest so much time that pivoting feels like failure. You perfect instead of ship. You build in isolation. Then you try to shoehorn software onto a customer.

By the time you have something to show, you're too invested to hear that nobody wants it, too exhausted to promote it properly, too deep to pivot. This pattern repeated for fifteen years. Always building outside work hours. Never able to go full time. Spotting opportunities. Building solutions. Failing to ship.

What's Different Now

Last Sunday changed something for me. Getting MemoryDial running in four hours made me realise the game has completely changed. I can spin up a prototype by Wednesday. Test an assumption before lunch. The prototype is the customer interview now. Building is talking.

When you can build in hours instead of months, “build it and they will come” becomes “build it and see who shows up.” The cost of being wrong dropped to almost zero. But more importantly, I finally understand what shipping means now.

The Blog Is the Shipping

MemoryDial is becoming two things.

First, the tool itself. I'll extend it however I want. Hook it into my AI systems. Make it do things Distill.io can't or won't. My own infrastructure. My own rules. I'll use it myself, and if others want to use it, great.

Second, MemoryDial.com is becoming this blog. A place to ship ideas at 80 percent instead of products at 100 percent. To write and think in public. To build momentum through publishing.

Writing is shipping now. Publishing is launching. Getting ideas out there. Seeing what resonates. Keeping things moving.

The technical details are genuinely fun. How Django's old URL routing needed updating. How WebSocket implementations have evolved. How what took six months of learning and building in 2011 takes four hours of prompting and debugging in 2025.

But the technical details aren't the point anymore. The point is momentum. Getting ideas into the world while they still have energy.

Building Within Reason

The milkman didn't need our perfect system. He needed to not get robbed collecting cash at 3am. Twenty euro a month for payment processing would have changed his life. He wouldn't have been in a position to get robbed if he was in bed asleep while his money was collecting automatically. That lesson was right there in my face of spider webs, if I'd been listening instead of building.

Fifteen years of this pattern. Spotting opportunities but not shipping them. Building in isolation then wondering why no one came. Working nights and weekends. Never able to make the jump to full time because nothing quite took off.

The spider webs are still there. Problems still exist at 3am like cobwebs. But now I can build something as quick as I can write the spec. See if anyone else gets their face full of spider webs. Move on if they don't.

Building within reason finally makes sense when reason means hours, not months.

I'm more mature now. More focused and excited about sharing ideas and things I'm working on. Not trying to build the perfect system in isolation anymore.

The blog is the shipping now. Ideas out there. Building in public. Thinking through writing.

Welcome to MemoryDial. Again.