Summary
Impact
Challenge
Approach
Solution
Outcomes
A deceptively simple publishing flow that defined how solutions would be structured, discovered, and matched across Halo.
I designed Partnering Listings, a self-serve way for researchers to make their work discoverable to industry R&D teams. The simple publishing flow established the structured data model that enabled AI-powered matchmaking and shaped Halo’s product roadmap.
Industry couldn't see what researchers were available to partner on, and researchers had no fast way to show it. For Halo, the recently launched Discover feed couldn't reliably match demand to the right innovators without structured supply. Listings were the supply layer the marketplace was missing.
Self-serve listings a researcher can publish in about three minutes, straight from their profile. On the other side, industry sponsors get a proprietary view of what researchers are ready to partner on right now, with structured signals they can't find elsewhere.
I owned product design from research and the PRD through flow design, interactive prototyping, testing, and launch. Working with the head of product, I developed the product model, explored alternatives, and brought recommendations to stakeholders. I used Bolt to build interactive prototypes and tested desirability and utility with Halo's user panel.
The full walkthrough, from the marketplace problem to the product model, the tradeoff that shaped it, and what that model unlocked.
At first glance, the goal could look like designing an effective form. However, the challenge for this feature was really designing the data model that the rest of the product would depend on.
A listing isn't a brochure, it's a structured signal for matchmaking. The hard part was a tension: it had to be open-ended enough to hold anything a researcher offers, a product, a method, a service, while still capturing enough structure for the feed to match on.
That made the model foundational, not cosmetic. Every surface a listing touches, the Discover feed, matching, messaging, and later facilities and lab listings, builds on the structure set here.
More information makes a listing more useful; more effort makes it less likely to get created. That tension set the three-minute goal. Three-step self-serve creation, straight from a researcher's profile, became the minimum viable behavior.
The last decision was the one that mattered most. We could have auto-generated listings from researchers' publications and let AI rewrite them for industry. I designed those flows, but chose not to ship them yet.
For a scientific audience, a confidently wrong listing erodes trust faster than a slow, correct one. The MVP kept researchers in control: they write and approve what represents their work. We deferred generating listings from publications. Converting existing researcher-authored proposals was a separate approach that shipped shortly after the MVP, with researchers reviewing and approving each resulting listing.
.svg)
Converting existing proposals produced 3,600+ opportunities that weren't available elsewhere.
Listings gave Discover structured information about researchers' capabilities, readiness, and partnership interests. This made research already on Halo available for ongoing discovery beyond individual RFP responses.
Richer content and featured placement gave innovators a way to stand out, and institutions pay for it.
What began as a three-step form became the structured supply the roadmap now hinges on. Listings shifted Halo's roadmap toward expanding supply: facilities and lab listings, and admin-assisted creation.
