AllCampus
Three and a half years with one client, from a CRM refresh to a platform direction, an AI career advisor and a rebuilt front door. It started with telling them the brief was wrong.
AllCampus helps working adults use the education benefits their employers already pay for. Over three and a half years it brought us three different problems. First, an internal CRM called Archa that gave advisors, admins and marketers the same screen for very different jobs. Then a learner platform that needed a first conversation to scale past one advisor's calendar. Last, a front door built for someone who already knew which program they wanted, when most people arriving didn't.
I owned design direction for the whole relationship, and the relationship itself on the design side. On Archa I did the research and the design myself. By the later phases I was directing: setting what design should solve, staffing it, and keeping the client's product and content leads inside each decision.
Inside Archa, advisors pieced together a student's status from memory and open browser tabs before calls where trust gets won or lost. As one enrollment advisor put it: "If I don't know the exact start date of a program, I can't even filter my prospects. I end up guessing or pulling reports from somewhere else."
The redesign
wasn't a UI problem.
Early designs proved a cleaner, role-aware interface was possible, and also that it wasn't buildable. The Identity Access Management model underneath couldn't deliver different experiences to different roles at all. Feasibility wasn't a checkbox here; it was the finding. Instead of treating it as somebody else's blocker, we made IAM reform the design recommendation and defined the solution with product and engineering. That decision unlocked role-aware design as a platform capability rather than a mockup.
Diagnose, then design.
The engagement started with people, not screens: stakeholder interviews across enrollment, marketing, admin, and leadership, mapped into a service blueprint of user journeys, backoffice systems, and internal handoffs. Three findings shaped everything. Advisors ran a high-volume, time-sensitive workflow without context. The system was rich in data but poor in visibility. And the platform treated all users equally when their jobs demanded the opposite. That research is the desirability case in this project: nobody had asked the people inside the tool what the tool was costing them.
Interactive prototypes became the alignment instrument. Simulating real tasks, filtering prospects, logging calls, navigating appointments, they gave siloed departments the same artifact to react to. Advisors saw their ideas land in the interface; directors aligned on priorities in front of it. "The prototype made me feel like, for the first time, the system was being built around how we actually work," an enrollment manager told us.
Alongside the structural work we shipped what the blueprint justified: a redesigned advisor dashboard, streamlined filters and tray views, standardized terminology, and a shareable bookmarking system. Targeted improvements, no total rebuild. That last clause was the viability constraint the whole engagement lived inside, and honoring it is why the recommendations got funded instead of filed.
Design with, not just for.
On Archa I mentored the client's early-career designer, guiding her through research synthesis, UI design, and stakeholder communication until she owned key deliverables and presented design work herself. That was deliberate capability-building on both sides: our team got stronger, and the client saw what design maturity grown from within looks like.
The engagement's value concentrated at two moments that had nothing to do with screens: recognizing that the IAM model, not the interface, was the real blocker, and choosing to grow a designer rather than just deliver around her. Both are direction decisions. The screens mattered, but they were the receipt, not the work.
A principle the AI couldn't break.
On Ally, the platform's AI career advisor, I owned the conversation design: the voice, the persona model, the pacing. The rule I set early was "a learner who feels interrogated will disengage." It held through every pivot the product went through, from alpha in March 2026 to beta in June, because it was written down before anyone was attached to a flow.
Build the spec, not the picture.
In 2026 the client's developer was rebuilding the learner experience in a no-code tool. A static design would've left him guessing at every state, so I built a working prototype he could click through and rebuild from, and treated the client's own content lead's words as the copy of record instead of rewriting them. When the client reversed a gating decision twice, the prototype took a toggle rather than a rebuild, because I'd kept both versions alive in the code.
20-30% more calls
Advisor call volume rose with clearer prospect data and workflows. Client-reported early outcome, measured post-launch.
25% faster onboarding
New advisors reached productivity a quarter faster on simplified navigation and a more focused UI. Client-reported early outcome.
Three and a half years
The CRM call earned the trust that carried the relationship through an AI advisor and a rebuilt front door.
Name who decides.
I'd name who decides earlier. On the 2026 phase we built for decisions that changed twice, and some of the inputs we were waiting on never arrived before the engagement ended. Keeping both versions alive saved us. Agreeing on one owner for each open decision on day one would've saved the week.