7.2.25
Building “Allies” With Canopy (And Falling a Little in Love With Lovable)
There are few things more fulfilling than a partnership that grows over time. We’ve been working with the team at Canopy for a few years now as design consultants, collaborators, and increasingly, trusted thought partners. They're kind, intentional, mission-driven people who actually care about the folks they serve. Naturally, we wanted to do right by them when a new opportunity came up.
A few months ago, Canopy approached us with an idea: what if their app had a built-in feature to support accountability partners—a way for someone to stand alongside a user and help them stick with their goals? They called this person an Ally.
The Idea
Canopy had already surveyed their users and found something pretty compelling: 65% of respondents said they already used an accountability partner in some form. That stat alone was enough to kickstart a deeper dive. So, our team at Goodfolks followed up by conducting user interviews with a handful of those people to understand the informal ways they were already doing this. Spoiler: it wasn’t elegant, but it was deeply human.
From that, we began sketching out what an actual Ally feature might look like. There would be two distinct user experiences: one for the primary user of the app, and one for the Ally.
Here’s the gist of what we aimed to build:
A simple way for users to invite someone to be their Ally.
An onboarding flow for the Ally once they accepted the invite (on a separate, standalone web page).
A dashboard where Allies could approve or deny protection downgrade requests, send encouragement, tweak notification settings, or gracefully exit if needed.
So, a tidy list. But building it? That’s never as straightforward as the bullet points make it seem.
Enter: Lovable
Canopy asked if we’d prototype the feature before heading into design. But instead of spinning up our usual toolset, they asked if we’d try a relatively new platform called Lovable—a prompt-to-code tool we’d been eyeing anyway. Naturally, we jumped at the chance.
For some context: I (the author) started my career 17 years ago as a front-end developer. These days I’m more in the UX and UI design world, but I still stay up-to-date on front-end tech, CSS tricks, and anything that scratches that “what-if-I-coded-this-myself” itch.
So when it came time to prompt Lovable, I… kind of went overboard.
My first prompt ended up being 1,445 words long.
[Prompt content would go here—obscured or collapsible to keep the post readable.]
Lovable handled it like a champ. I even threw in a request to use SilkHQ’s modal and bottom sheet interactions—just linked the site and said “make it like this.” It did.
(Quick side note: this was purely a prototype for internal testing, not a production build. If this had gone live, we would’ve fully licensed SilkHQ’s UI kit. Gotta support the good folks making good stuff.)
The Good, the Buggy, and the Beautiful
Was the Lovable output perfect? Definitely not. But it was shockingly good. Enough to fully communicate the concept, enough to share, test, and iterate. I even ran the same prompt through Vercel’s v0 for fun, and Lovable still came out on top—especially in terms of UI quality.
I made a few small tweaks manually—synced to GitHub, pulled locally, and edited directly. That was the best of both worlds: I used Lovable for the heavy lifting, and then my own CSS brain for the polish.
One of the coolest things? When we needed to support multiple users (i.e., one Ally for multiple people), I prompted Lovable and it just did it. No hours spent Googling how to manage selectors in React. It Just Worked™.
And then came a magical little moment: Lovable noticed the file was getting long and asked, “Should I refactor this into smaller files?”
Yes, Lovable. Yes you should.
Things Got Real
We presented the prototype to Canopy and the response was fantastic. Naturally, they had some new thoughts after seeing it in action (this is why we prototype, after all). These updates meant reworking a chunk of the Ally dashboard.
That’s when I hit my first real snag with Lovable: it did what I asked… but it also broke the code. Compilation errors, sadness, confusion.
Thankfully, I had a backup. I pulled the changes locally, diagnosed what I could, and for the rest? I dropped the errors into ChatGPT. It calmly told me what was wrong, and even how to fix it—without judgment. Honestly, that back-and-forth between tools made me feel more capable, not less. It was collaborative, not adversarial.
Where We Landed
Eventually, we arrived at the final prototype, presented again, and saw another great response. But after internal deliberation, Canopy made a strategic decision: for an MVP, they’d strip things back. No accounts or dashboards for Allies. Instead, each request would come via a one-time link with a simple approve/deny interface.
It was the right move. And it reinforced a big learning: you don’t always know what’s essential until you build something real and test it with real people.
That’s why I’m so excited about Lovable. It let us build something real in just a couple days. Not just wires. Not mockups. Not even a Figma prototype. But working code. Clickable. Shareable. Breakable. Fixable.
And because it was fast, we could afford to get it wrong, regroup, and get it right.
Where We Go From Here
We’re continuing to work with Canopy to refine the Ally concept and see how it evolves. Lovable was a fun tool to explore, but the real joy came from helping a team we respect move one step closer to serving their users better.
Want to Prototype Something Together?
We’d love to help you do the same—whether it’s a messy idea you haven’t quite figured out, or a rough feature you’re itching to test. No pitch decks, no polished fronts. Just good people, making good things, with a little curiosity and a lot of heart.
