Chimp

Chimp was built for two types of users: students trying to figure out their direction, and professionals actively working on their growth. Both were trying to bridge the gap between where they were and where they wanted to be — they just didn't always know what that gap looked like.

Every decision was a hypothesis. Research helped — studying how people use LinkedIn, Coursera and similar platforms gave us signals — but there was no way to test assumptions until the product was in front of someone real.

The founder came with a vision; I came with a blank canvas. Most core decisions — how AI-generated content should be presented, what a roadmap state should feel like, how to surface personalisation without making it feel algorithmic — were ours to make from scratch.

Home, Roadmaps, Profile, Connect, Chat — each was its own challenge, on both web and mobile. Keeping the experience consistent across all of them, with a partial design system and a tight timeline, was its own problem to solve.

The Home screen is where everything comes together. On the left, a summary of your progress — conversations with Chimp, advancement percentage, networking activity. In the centre, a personalised feed filtered by jobs, courses, contacts, events, news and Chimp-generated content. On the right, the Quickchat panel — the AI is always a question away, regardless of where you are in the product.

Designing this screen meant understanding the hierarchy before touching layout. The feed is the main event; the summary and Quickchat are supporting context. Getting that balance wrong — giving too much weight to either side — would have made the screen feel like a dashboard rather than a starting point.

The profile has two jobs. It gives the AI the structured data it needs to personalise your roadmap and feed — goals, skills, experience. And it presents that information to other users: potential mentors, collaborators, employers.

That double job created a real tension. A profile optimised for machine readability is a form. A profile optimised for human readers needs to feel like a person. The challenge was finding a layout that worked for both without feeling like either.

I focused on the CV layout — the most recognisable format for the professional audience Chimp is targeting. When a roadmap is generated based on a profile that looks familiar and complete, users are more likely to trust what comes out the other end. The AI's input fields sit in a secondary section — they matter to the system, but they don't need to dominate the human-facing view.

Users can have a personal profile, a mentor profile, or both.

Roadmaps tell you what to learn. Connect is about who to learn alongside. It surfaces people worth knowing — professionals and experts the AI recommends based on your profile — alongside events in your field and what your existing connections are already attending.

This was the screen that changed most between first sketch and final design. I show that process here because the iteration is part of the story — and because Connect is also where the Chimp Quickchat panel is most clearly visible in context. The AI isn't just behind the roadmap; it's available throughout the product.

The sketch showed a grid of connection avatars on the bottom left — who you're connected with, nothing more. The final design replaced this with social signal cards: "95% of your connections are attending the seminar with Dr. Sturgis."

A list is passive information. "80% of your connections are going to this event" is a signal — should you be there too? This also reflects what Chimp's AI is actually for: not just surfacing data, but extracting meaning from it. The change made the panel actionable rather than decorative.

In the sketch, three separate ideas sat in three separate columns — connections on the left, events in the centre, people on the right — with no clear relationship between them. Users had no obvious starting point.

In the final design, the centre became entirely dedicated to discovery: people first, then events. Both serve the same user intent — find someone new, or join something new. Grouping them together means a user can answer "what can I do here?" in a single glance.

In the sketch, everyone appeared in the same list. The final design introduced two tabs.

The reason: someone at your level who you might collaborate with is a different kind of connection from someone more experienced who you can learn from. Mixing them in one list blurs both — the user doesn't know who they're looking for. The tab split lets users decide their intent first, then browse accordingly.

Chimp recommends roadmaps based on your profile — your goals, skills and interests. Each roadmap shows enough information to help you decide whether it's right for you before you commit: the title, a short description, and how long it takes to complete.

Chimp's AI isn't confined to the roadmap. It's present across the entire product — on web, as a panel on the right side of every screen, surfacing questions relevant to whatever the user is looking at. On the Roadmap it asks about your progress; on Connect it asks about events and people. The context changes, the AI stays.

On mobile, it lives as a persistent bar at the bottom of every screen. One tap opens a full conversation — where users can ask anything, upload documents, or pull in content Chimp has already generated for them.

Each feature started with a conversation. The founder described the idea — sometimes in detail, sometimes as a direction — and I'd ask questions until I understood what problem it was solving and for whom. "It would be great if Chimp could give users a roadmap" is a starting point, not a brief. Asking what a user would do with it the first time they opened it turned it into one.

Without users to interview, research was how I grounded decisions. I looked at how comparable products — learning platforms, professional networks, AI assistants — handled the same challenges. Not to copy them, but to understand what conventions already existed and where there was room to do something different.

For complex screens, I worked out structure on paper before opening Figma. It's faster to rearrange columns with a pencil than with frames. The Connect wireframe is the most visible example — the layout decision that mattered most was made before a pixel was placed.

Designs were shared with the founder and team before anything moved to development. Most of the changes that matter — the roadmap's before-start state, the Connect layout revision, the profile's treatment of AI-facing fields — came out of these conversations.

As a first-time designer, I sometimes focused more on how a screen looked than how easy it was to use. On mobile, I moved the Profile Configurations panel behind a small arrow on the side of the screen because it made the main view look cleaner. But it also made a core part of the profile much harder to find — a user who doesn't notice the arrow will struggle to reach it. Looking back, I would have spent more time asking whether users could do what they needed to do, alongside thinking about how the screen looked.

As a first-time designer working at speed, I often jumped straight into designing a screen without fully thinking through how it connected to the rest of the product. The missing connections list in Connect is a good example — I focused on making the screen cleaner and more useful, but didn't step back to ask whether a user could still do everything they needed to do. Slowing down at the start to map out the full user journey would have caught gaps like this before they made it into the final design.