Journal
Building Archie for home projects
Archie’s beta has made it clearer who I’m building for: people with a room to rethink, a renovation to plan, or a home to make their own. Here’s what I’m learning from them.
The beta at a glance
- 65%
- of people who signed up got their first designs within a week
- 38%
- of them showed signs of a serious project: a concrete need, repeated iteration, and coming back or moving towards implementation
- 28%
- of them took a step towards implementation inside Archie: the gap I’m working on next
Archie started from my own problem: deciding what to do with your home is hard. I first had the idea after a leak in our kitchen turned into a serious rebuild and redesign. Many decisions, big budgets, stressful calls, and no idea how to pick a tile.
Most AI design tools and image models are great at giving you an inspiring picture. But that only gets you so far when you have an actual project to work through.
The open beta was a chance to see who gets value from Archie as it exists today. Let’s call it Archie v1.
My main takeaway: there’s a meaningful group of people with serious home projects who get more out of Archie than out of free or general AI tools. If I can solve their needs well enough, there’s a paid model and a go-to-market motion that can turn this into a real business.
01The use cases
Most people came with a project, not just looking for inspiration.
I analyzed every room brief people wrote and grouped them by the job behind it. Four use cases emerged: whole-room design, specific decisions, renovations, and general inspiration.
General inspiration is the use case I want to avoid building around. In my experience, ChatGPT, Gemini, and a whole range of AI image apps already handle this well. Take a picture of your room and ask them to make it better, and they can do a solid job.
Where I find they fall short is on projects that need more depth and ongoing work. So I was happy to see that about four in five activated users came with something more concrete: furniture, paint and layout decisions that all affect each other; moving a wall to make a room work differently; revamping a kitchen or bathroom.
02How people used Archie
More rooms and more designs point to a real project.
Rooms worked on × designs generated · each dot represents one or more people
- Whole-room redesign
- Specific decisions
- Renovation
- General inspiration
26%Tried one room1 room, up to 4 designs
22%Worked on one room1 room, 5 or more designs
30%Took on two rooms2 rooms, 6 or more designs
22%Worked across the home3+ rooms
Once I understood what people asked for, I looked at what they did next. Two simple metrics do the heavy lifting here: how many rooms they worked on, and how many designs they went through.
About a quarter tried one room and stopped after a few designs. Everyone else kept going: working a single room through, taking on a second, or working across three rooms or more. That last group is who I think of as whole-home users, and their projects are the ones I’m most interested in understanding.
03Engagement
Concrete projects led to more engaged usage.
Share of each group · the tick marks the beta average
Whole-room redesign
“Make our living room feel elevated and work for a family with a toddler. Happy to replace furniture.”
Specific decisions
“What is the biggest comfortable couch we can actually fit here?”
Renovation
“Remove the bathtub, add two showers and swap in a marble-effect tile.”
General inspiration
“Make it warmer.”
- Whole-room redesigns and specific decisions both brought people back to iterate, and more than a third of each group took a step towards implementation. These are the kinds of use cases I built Archie for.
- Renovators also iterated, but fewer took an implementation step. Quotes, contractors and permits may play a role. Candidly, I haven’t fully built for this phase yet, and it’s on my list.
- General inspiration showed less iteration and fewer implementation steps. That was expected: if all you wanted was a pretty picture, you might not need much else once you have it.
04A note on prompts
Intent and projects matter more than prompt length.
Whole-home project
“Pimp my house please.”
Came back across several weeks, room after room
Whole-room redesign
Brief summaryMake our living room feel elevated and functional for a family with a young kid. We’re happy to replace the furniture.
Kept iterating on sofas, rugs and colour
General inspiration
“Make this a lot nicer.”
Stopped at the first designs
A detailed brief can be a useful signal, but it doesn’t tell the whole story. Some of the most engaged users started with just a few words. What mattered was the real project or need behind them.
That’s useful for how I think about Archie’s first questions. I want to understand what someone is trying to do, without expecting them to write a perfect prompt to get started.
05Real homes
People are making real homes work.
- “Where I live, sleep and cook”
- “A guest bedroom that is also a nursery”
- “A shed becoming a home office and gym”
- “A living room with a play area”
This surprised me most. People are working around small spaces, changing family needs and rooms that have to do several jobs. These are homes people live in every day.
A lot of my early work on Archie focused on making spaces more beautiful and well considered: bringing some of what an interior designer brings, while preventing image models from inventing walls or changing the size of a room.
But the needs people brought went well beyond how a space looks. They needed a living room that works for kids, somewhere for guests to sleep, or a small space that can do three jobs. What a room has to do matters as much as how it looks.
We had previously been using Gemini or ChatGPT … it always invented windows or walls, or made the space bigger than it actually was. I loveeee the ability to keep certain pieces of furniture so I can see what I actually need to budget for and not an entire new space.
06The strongest signal
People who started with several rooms were the most likely to come back.
One room
68% of users
32%came back another day
Two rooms
20% of users
20%came back another day
Three or more rooms
12% of users
67%came back another day
This was the clearest signal in the beta. Working on more rooms takes more visits, so to keep this fair I looked only at how many rooms people set up on their first day, before they had any reason to return. People who started with three or more rooms were about twice as likely to come back as people who started with one. Two rooms on day one didn’t make a difference on its own.
The scope of the problem changes how much someone might need a tool like Archie.
In the same few weeks, one person might plan a bathroom renovation, pick a paint colour for the stairs and furnish a child’s bedroom. It looks less like a design session and more like a running list of house projects. That’s the kind of ongoing need I want to build for.
07The gap
Coming back to Archie isn’t the same as getting your project done.
Among people who worked on three or more rooms
People working across three or more rooms kept returning, yet only about a third took a step towards implementation inside Archie. There’s a clear gap for me to work on here: helping someone choose a piece, source it, or prepare a change to their home.
Those actions are useful signs of progress, but they don’t tell me whether someone bought the sofa or finished the renovation. I want to get better at helping people make that next step, and understanding what happens after they leave Archie.
The shopping list functionality is really a massive plus! Is there a way to select your country of residence to find more local shops?
08What’s next
Building Archie for home projects.
Focused projects
The room project
A room to rethink or a specific decision to make. These are the projects I built Archie for, and where I’m seeing people take steps towards implementation.
52% of activated users
Ongoing home projects
The whole-home project
Many decisions across one home, worked through over weeks. This is the ongoing need I want to build for as I work out the product and paid model.
22% of activated users · 82% came back
The best part of launching an open beta has been talking with so many homeowners and people with real home projects. It’s clear there’s a real problem here, and that existing tools aren’t solving enough of it.
The immediate next step is to figure out a paid model that delivers value for these projects, and a way to reach the right people profitably.
Beyond that, a few things are top of mind, and I’ll share more as I go:
- Making it easier to move from an idea to implementation.
- Making Archie more proactive, and more of an agent for your home.
- Thinking about the full homeowner journey and what triggers a project: a leak that turns into a renovation, a growing family, or selling your home to upsize or move neighbourhoods.
That kitchen leak was my starting point. I’m learning how many different starting points there are for everyone else.
Have a home project in mind? Check out Archie at usearchie.com.