The Honest Wrap-Up

The last one

Fifteen weeks. Two tools. A lot of bugs, a lot of rebuilt pipelines, and a few moments that made all of it worth it. This is the wrap-up — honest, not polished.

What Alex would tell someone starting here

There’s a version of a co-op where you show up, complete tasks, and leave. This wasn’t that.

The difference between school projects and real work isn’t the complexity — it’s the feedback loop. In school, you submit something and wait for a grade. Here, you build something, put it in front of a real person, watch them use it, and then go fix what didn’t work. That cycle — implement, review, test, repeat — is the actual job. Getting comfortable in it early is probably the most useful thing this placement gave me.

The hardest part of the last week was the most instructive. A storage conflict between tools caused bugs that could only be identified after a full brief was built and a site was generated. There was no shortcut to finding them — the only way through was to run the whole pipeline and see what broke. Debugging that way is slow and frustrating, and it’s also exactly what production work looks like.

If I started again, I’d skip the tools that weren’t ready — tools like Stitch that looked promising in demos and fell apart under real conditions — and get to what actually worked faster. I’d also outline the seven-step wizard model earlier, because by the end it was clearly the right structure. We arrived at it through iteration, but knowing what I know now, I’d have started there. And I’d push earlier to experiment with having the build process run inside the tool itself — right now a developer has to pull the output into a codebase to see a real prototype, and removing that step would open things up significantly.

The thing I’ll take with me is less about the tools and more about the people. Meeting designers, social teams, account leads — people doing the actual work, with actual opinions about their workflows — is a completely different kind of learning than anything that happens in a classroom. Their feedback shaped both projects in ways that no amount of internal testing would have. AI is a useful collaborator. People who do the work every day are irreplaceable.

What Paige would tell someone starting here

The hardest stretch was near the end. Real ads were being produced through the tool for real clients — actual deliverables, on a deadline — when an unknown storage conflict between the two tools surfaced and required redeployment to fix. The timing was the hard part. When the stakes are real and the tool needs work at the same time, the pressure is different from anything a school project prepares you for.

The most satisfying moment was also near the end. Seeing ads that came out of the tool — ads that looked great, that clients approved, that went live — show up on my own social feed. Something built over fifteen weeks, used by real people, in the wild. That doesn’t happen in school.

If I started again knowing what I know now, I’d spend more time upfront mapping out the small things — the UI adjustments, the workflow changes, the features that make the tool easier for the people using it. Not because those things are more important than the bigger features, but because getting them out of the way early frees up the second half for expanding what the tool can actually do, rather than modifying how it works for users. The scope of those smaller changes only becomes clear once you’re in it — but if I could see it from day one, I’d tackle it first.

For Alex, this co-op is a final look at what’s coming — he’s heading into fourth year and then into the workforce, and this placement was a preview of that. For me, it was something different. I’m earlier in school, and this was a glimpse well ahead of schedule — of what work actually feels like, what skills actually matter, and what I want to keep building toward before I get there. I’m taking that with me through the rest of it.

What both of us would say to someone applying next summer

This kind of co-op gives you a lot of room to take your project somewhere. That’s the opportunity and the responsibility at the same time.

Plan early. Ask questions. Ask for feedback more than feels comfortable. And really listen to your users — not just when they tell you something is broken, but when they describe how they work and what would make it easier. It doesn’t matter if you don’t personally see the value in a feature they’re asking for. If it makes their workflow better, the tool gets better. User experience is built from listening, not just from building.

There’s so much to learn from people who are already in the field doing the work. AI is a genuinely useful tool. But the people around you — the designers, the strategists, the account leads, the clients — are still the best resource in the room.

We had a great time. Thanks for following along with our two AI projects!