- A calmer, three-step conversation with talent. The outreach assistant now talks to a freelancer in three clear steps: it makes a decision, walks through the key checks — scope, then budget, then timeline — one at a time while just collecting answers (not judging as it goes), then asks any remaining extras together in a single message. It waits for a proper review before it replies, and if the work simply can't be scoped it now declines cleanly and kindly — all in a warmer, Mira-style voice.
- A new teammate writes each match's "why they fit." A new behind-the-scenes agent — PORTIA, the fifteenth — now writes a short, human pitch for every finalist at the moment they're revealed, grounded in that person's real skills and only their genuinely standout stats. It replaces the old bland bullet list and sits in a bigger, roomier card.
- Search leads with what someone can actually do. The talent scout now weighs a freelancer's proven capability first, ahead of softer preferences, and shows a clear quality grade — on both the guided search and the self-serve one.
- Talent's own files get read and scored. When a freelancer attaches files, the assistant now reads them straight into the conversation and the offer-scorer weighs them — the same pipeline that reads your uploads. Files from talent are trusted and handled inline.
- The results cards get a full redesign. The match cards were reworked to the new design frame: compact cards, each person's details tucked neatly under their name and photo, a "search on your own" pill, portraits that never get squeezed out of shape, and an accessibility polish pass across the results.
- Smoother brief & question moments. The brief checklist gains a search-preferences row and auto-opens itself when it's time to review, answering a concierge question earns a small thank-you, and picking or comparing more than one candidate now badges cleanly.
- Money and honesty fixes. The card shows a real hourly rate instead of a package price mislabeled as one, budgets are always committed in dollars, a skipped budget or timeline reliably triggers a fresh estimate, and empty research now leaves a clear note rather than going silent.
- Sturdier under the hood. The database now rides out a start-up blip instead of dying, the system measures how often the AI reuses cached context (cheaper and faster), and a failed sign-in send logs the client out cleanly rather than silently failing again.
The heaviest day of the week: the talent conversation is rebuilt into three calm blocks with a real review before any reply, PORTIA joins as the fifteenth agent to write each match's human pitch, search leads with proven capability, talent's own files are read and scored, and the crew lands a full match-card redesign plus a raft of correctness, brief-checklist and resilience fixes.