Hack Week 1: what an inspiring bunch

Real projects, real messiness, real progress

We’d piloted the format before, but this was the first time running it for real: a full group of paying participants, a clear structure, and the big ambition of helping non-engineers explore how AI can unlock better ways of working.

We couldn’t have asked for a better group.

Curious, open-minded, and collaborative, each person showed up ready to explore this new territory together. There were moments of excitement and momentum. And yes, plenty of moments where things broke, didn’t make sense, or hit a wall. That’s part of the process.

We came into this week with the belief that using AI tools is really the only way to get a grip on how they can be leveraged. But as designers, we also knew that just because you can build something, that doesn’t mean it’s going to work.

And work it didn’t. Pretty much everyone in the cohort experienced the trough of disillusionment at one point, as we both did as participants before them.

The AI trough of disillusionment from our intro session

But pushing through, thinking critically about where to use AI and which pitfalls to avoid (as well as the accountability and conversation from the group) everyone ended up invigorated - if not by what they built then at least by the new knowledge and understanding they gained.

So what did we build?

From our initial calls we knew we had a diverse group of participants with different problems to solve. It was fascinating to watch them work through those problems and discuss them together.

Marketing tools

Paul’s marketing copy generator

Two of our participants, Paul and Nick, were coming from a marketing angle. Marketing tools like Jasper were the first to emerge from the initial LLM boom before even chatGPT was on the scene, so both were very familiar with the potential of AI to help with writing. But custom workflows built specicifically for their tools, editorial tones and toolkits had been elusive before the Hack week.

Within a few hours both had identified opportunities for tools and got building. Using existing scripts and prompts from their work, they created bespoke marketing tools in replit quickly, then discovered the challenges of iterating on vibe-coded tools for themselves (over-enthusiastic AI breaking one thing as it fixes another.)

Nick built a social media inspiration tool specific to his vertical, and Paul built an AI copywriter capable of keeping to his style andcontext over articles of more than 500 words (a real challenge for most AI tools)

AI accountability buddy

Nori, Dessy’s cosy focus companion during her demo

Dessy tackled a personal pain point with an AI accountability buddy to help her keep on track with whatever task she was focused on. Kind of an ADHD support app, which ended up with a seriously cute cat mascot.

Dessy’s week perfectly encapsulated the challenge with building with AI. You still need a problem and solution that speaks to the problem. Dessy built a functioning pomodoro timer wrapped in a cute app with gentle encouragement and a chat functionality, which failed to solve her problem.

Sometimes only by building something can you find out that it doesn’t work for you. Regardless it was a lovely project which we had a lot of fun building a UI and brand for.

Slide summariser

Deckster in action

Darvish, with a strategy background, was very keen to simplify a workflow he finds himself in often, to generate slide decks with the right visual and content style for his business and board.

His product “Deckster” (Great name!) had the goal of being a complete presentation generation platform. Like many in the group, Darvish experienced frustration early on attempting a custom GPT approach before opening up Replit and attempting to vibe code his solution.

While Darvish admitted he might have structured his week differently, he was amazed at how quickly he could tick off the first feature he needed - an instant summary of PDF presentations he was uploading. While he never got to the ultimate vision, this was already an unlock not just literally but mentally.

One on One voice assistant

Paula’s voice assistant

Voice based AI experiences are getting incredibly powerful, and Paula dove into the world of voice with a one on one voice assistant.

Paula thought she was going to build something completely different, but with a stated aim of ‘just exploring’ she found a more interesting problem to solve technically during our initial workshop: a way to do a walking 1:1 even if there’s no-one around to walk with’.

From there she was straight into voice mode, with constant prototypes we all piled in to test every morning, and a rapid iteration cycle. Paula also really leant into the final day branding and UX session and ended up with a wicked retro UI. Check out her demo at spokenx.com.

Beyond the hacks

More important than the individual projects was the broader exploration and group learning. Our cohort group chat is still busy with people’s ongoing work. Two cohort members have even joined the first ever Lovable Shipped cohort.

There was plenty of frustration during the week - dead ends hit, tools that didn’t work, infuriating sychophantic AI assistants. But doing it together in a small group, talking for over an hour a day and sharing challenges meant where each of us might have given up ordinarily, we pushed on.

Onwards

We loved running the first hack week, and we’re already planning the next one. If you’d like to join the next hack week, sign up below. We’ll see you there.