An HR Screening Assistant

The problem

First-round candidate screening is repetitive, time-consuming for recruiters, and inconsistent between interviewers — the same role gets a different bar depending on who happens to run the call. An HR technology company wanted a consistent, always-available first-round interview that produces a structured, auditable score rather than a recruiter’s unrecorded gut read.

The integration

The interview itself is a Wetel workflow: a persona configured as an interviewer, a multi-turn screening conversation (typically 2-4 questions, confidence-gated so a low-confidence answer triggers a clarifying follow-up rather than a hard pass/fail), and an end-of-session evaluation that produces a structured score and rationale rather than a freeform transcript a human has to re-read in full.

The distinguishing piece of this integration is the avatar layer: rather than a text-only chat interface, candidates talk to a real-time digital-human avatar. Wetel’s IAvatarRenderBackend abstraction (see Avatar Config) let this customer plug in a third-party avatar rendering vendor as a swap-in backend, without touching the conversation logic underneath it — the same workflow and scoring pipeline works whether the frontend renders a browser-native avatar or a third-party photorealistic one.

Why the avatar abstraction mattered

Most “AI interviewer” products bundle a specific avatar vendor into the product itself, which means switching vendors later means rebuilding the integration. Because the avatar rendering backend is a pluggable abstraction rather than a hardcoded dependency, this customer’s actual product decision — which avatar vendor to use, and whether to change that later — never touches their conversation logic, scoring rubric, or session data at all.

Result

A consistent, scored first-round screening interview running unattended, with the underlying avatar rendering vendor decoupled from the conversation and evaluation logic — a decision this customer can revisit without a re-integration.