Book a call
← All posts

Wanderly: the AI layer for a 663,000-nurse staffing marketplace

Wanderly case study: Curate campaigns, CoPilot AI recruiter and a candidate Readiness Score built on a live healthcare staffing marketplace

Wanderly is a travel healthcare staffing marketplace. Nurses and allied health professionals find contracts on it; 500+ staffing agencies recruit on it. It's been running since 2017 out of Miami and has 663,000+ registered users. In 2025 it was acquired, and the new owners wanted it to do more than list jobs.

We've been the product and engineering partner for that.

What they came with

A marketplace that worked, and a recruiting process around it that was mostly manual. Recruiters searched, messaged, chased documents and scheduled by hand. Candidates filled in half a profile and went quiet. Agencies had no way to run a campaign to their own candidates from inside the platform, and no way to tell which of their contacts were actually ready to move.

The constraint that shaped everything: one team, the existing product had to stay live, and nothing could be built in parallel. Every phase had to ship something real.

What we built

Curate. A campaign platform for agencies, embedded inside the recruiter portal rather than sold as a separate tool. Job-first campaigns over email, SMS and in-app, with audience targeting, templates with job slots, sender identities on the agency's own domains, compliance limits, and campaign analytics.

CoPilot. The AI conversational layer that sits over SMS, email, web chat and the app. It starts the conversation, asks qualification questions, collects documents, and hands off to a human recruiter in the same thread when the candidate asks for one or when it hits something it can't answer. The handoff is a designed thing, not a fallback: the recruiter gets the candidate's exact words, what CoPilot already tried, and one deep link.

Readiness Score. A number on every candidate that tells a recruiter who to call first. Version one uses three signals: campaign engagement (50%), application activity (30%), profile completeness (20%), with recency decay and hard overrides. Opt-out forces the score to zero and removes the candidate from every campaign. Twelve months of inactivity caps it at 25. Tiers are Hot, Warm, Cold, Inactive. The weights are locked; admins configure the time windows and thresholds.

Recruit by Wanderly. The recruiter-facing mobile app: pipeline, inbox, candidates, dashboard, with CoPilot and Curate surfaced inside it.

Wanderly 2.0. The rewrite underneath all of this, phased so the old product never went dark.

The decisions that mattered

  • Curate lives inside the recruiter portal. A standalone tool would have been easier to build and nobody would have opened it.
  • The AI hands off to a human in the same conversation. Candidates in healthcare staffing don't want a chatbot; they want a recruiter who already knows what they said.
  • The Readiness Score went live with three signals, not eight. The full model existed on paper. Shipping the simple version to a real agency and learning from sixty days of data beat shipping the clever one late.
  • Opt-out is a hard override, everywhere. Compliance is not a score input; it's a switch.

Why this matters if you're building something similar

Most of this wasn't hard engineering. It was sequencing: what ships first, what stays manual, where the human sits. That's what the 14-day Blueprint is for. You get the prototype, the architecture, and the order of operations before anyone writes production code. Here's what one looks like: decipheringlogic.com/roadmap-sample

Have a feature stuck at the demo stage?

The $5,000 Decipher Blueprint de-risks it in two weeks — architecture, scope, and an honest quote. Credited if you build with us.

How the Blueprint works