HelloCraft AI: voice-first first-round candidate screening, powered by Vocal Bridge

An existing autonomous recruiting framework adds a Vocal Bridge voice agent, Senna, to run first-round screening calls and pass forward only qualified candidates.

The problem with first-round screening

First-round screening is the part of hiring that scales worst. A recruiter posts a role, candidates pour in, and someone has to talk to enough of them to separate the qualified from the merely interested. That work is repetitive, it is bottlenecked on the recruiter's calendar, and it is the reason strong candidates sit untouched in a queue for days.

The HelloCraft team's bet is that the screening conversation itself, the part where you confirm a candidate actually has the experience the role needs, can be handled by an agent without losing the feel of a real interview. Questionnaires get abandoned and they flatten everyone into the same boxes. A short spoken screen does the opposite. It adapts, it follows up, and it lets the candidate ask their own questions.

What HelloCraft AI built

HelloCraft AI had already automated the front of the recruiting funnel, importing a role, discovering candidates from LinkedIn and internal databases, scoring them against the job, and auto-advancing the qualified ones. The hackathon work gave that pipeline a voice for the step that matters most, the first-round screen.

Instead of sending qualified candidates a form, HelloCraft calls them with Senna, a voice agent that asks role-specific questions, evaluates the answers, and only advances the people who clear the bar. The recruiter gets a shortlist instead of an inbox. Voice fits because a screening call surfaces what a questionnaire cannot, on the candidate's own schedule.

How HelloCraft AI uses Vocal Bridge

HelloCraft AI is built on Vocal Bridge's Voice for your agent surface, where voice is layered onto an autonomous workflow rather than a hand-driven UI. The recruiting framework owns the sourcing, the candidate scoring, the internal database, and the outreach. Vocal Bridge owns the voice itself, the browser-based call, the streaming text-to-speech, barge-in so a candidate can interrupt with a question, and the sub-second turn-taking that keeps the screen feeling like a conversation instead of a phone tree.

The architectural split is clean. HelloCraft decides who to call and what to ask. Vocal Bridge runs the call.

from vocalbridge import VoiceAgent

senna = VoiceAgent(
    name="Senna",
    instructions=(
        "You are a recruiter screening for the role. Ask the role-specific "
        "questions, evaluate each answer, and answer the candidate's questions."
    ),
)

@senna.on_call_complete
def review(transcript, candidate_id):
    result = evaluate_responses(transcript)        # HelloCraft's scorer
    if result.qualified:
        advance_candidate(candidate_id, result)    # forward to the hiring team

Illustrative. The production framework wires richer scoring and candidate state into the pipeline.

What's hard about voice here

A screening call has to handle the candidate going off-script. People interrupt with their own questions about compensation or the team before finishing the prior answer, and an agent that talks over them or stalls stops feeling like an interview. Vocal Bridge's barge-in and low-latency channel let Senna be interrupted and recover cleanly.

That recovery is what lets HelloCraft hand a genuinely conversational first-round screen to someone who never gets to meet a recruiter first.

"Hello, I'm Senna, your recruiter for this role. I'll be walking you through a short set of questions to assess next steps, and I'm here to help answer any role-related questions you may have." Senna, HelloCraft AI's voice recruiter