BrewVoice

BrewVoice is SkillBrew.AI's AI-powered voice screening agent, the product that conducts automated, structured telephone screening conversations with job candidates at any volume, at any hour, in multiple languages, and without requiring a human recruiter to be on the call. BrewVoice is the front-line screening layer of SkillBrew.AI's hiring platform: the first structured point of contact between the organization and an applicant after their application is received.

When a candidate submits an application and passes initial ATS filtering, BrewVoice can be triggered automatically, calling the candidate at their preferred time, conducting a structured screening conversation, analyzing their responses in real time, and generating a complete screening summary that is immediately available for recruiter review. The entire process, from trigger to output, requires zero recruiter involvement for routine screening conversations.

BrewVoice is not a phone tree or an IVR system. It is a conversational AI voice agent, one that speaks naturally, listens actively, asks follow-up probes where appropriate, and generates substantive, nuanced screening output rather than collecting binary yes/no data.

The recruiter phone screen is one of the most time-consuming and least scalable activities in the hiring workflow. For every candidate who passes initial ATS filtering, a recruiter needs to:

• Email the candidate to schedule a call
• Find a mutually available time slot
• Conduct a 15–20 minute structured conversation
• Take notes during the call
• Write up screening notes and enter them in the ATS
• Make a pass/advance decision
• Send appropriate follow-up communication

At 20 candidates per role, this is manageable. At 200 candidates per role, it is operationally impossible at quality without either sacrificing the process or dedicating all recruiter time to it. At 2,000 candidates across 50 roles, it is simply not possible.

BrewVoice removes this constraint. Every candidate who qualifies for a screening call gets one, immediately, consistently, at scale, with no recruiter coordination required.

Step 1: Candidate Invitation

When a candidate is selected for screening, either by passing ATS filters or being moved forward manually by a recruiter, BrewVoice sends an automated invitation. The invitation explains:

• That the screening will be conducted via an AI voice call
• What the call will cover and approximately how long it will take
• How to schedule the call (candidate selects from available time slots, or requests an immediate on-demand call)
• What happens with the call output and who reviews it

This transparency is non-negotiable, candidates should know they are speaking to an AI screening system, not a human recruiter.

Step 2: The Screening Call

At the scheduled time, BrewVoice initiates the call. The candidate answers their phone; BrewVoice opens the conversation, introduces itself clearly as an AI screening agent, confirms the candidate's name and the role being discussed, and begins the structured screening sequence.

The screening script is configured for each role and organization, covering the specific eligibility and qualification criteria most relevant to that hiring context. A typical BrewVoice call covers:

• Availability confirmation: Notice period, earliest start date, current employment status
• Location and work model: Current location, relocation or remote flexibility
• Work authorization: Right to work in the relevant jurisdiction
• Compensation alignment: Current and expected compensation relative to the role's range
• Core eligibility validation: Specific requirements for the role, years of experience in a domain, possession of a required certification, language proficiency
• Role-relevant experience questions: 2–3 structured questions that probe basic competency relevant to the role, not a full interview, but enough to generate initial competency signal
• Motivation and interest: Why the candidate is looking and what draws them to this opportunity
• Logistics and next steps: Closing the call by informing the candidate about what happens next and when they can expect to hear back

The conversation is natural, designed to feel like speaking with a structured, professional interviewer rather than navigating a phone menu. BrewVoice uses neural text-to-speech voice synthesis built on Google Cloud's Chirp3-HD models, producing voice quality that is clear, natural, and appropriately professional.

Step 3: Real-Time Response Analysis

As the candidate speaks, BrewAI, the intelligence layer powering BrewVoice, processes every response:

• Speech-to-text transcription of spoken responses
• NLP analysis of response content against the screening criteria
• Communication quality assessment: clarity, coherence, language register
• Eligibility determination: does this response indicate the candidate meets the criterion?
• Competency signal generation for structured screening questions

Step 4: Screening Summary Generation

When the call concludes, BrewVoice generates a complete screening output, typically available within minutes of call completion:

• Structured eligibility data: Availability date, notice period, compensation expectations, work authorization status, all as structured, searchable ATS fields
• Communication quality score: An assessed rating of the candidate's verbal communication clarity and professionalism
• Role-specific screening question performance: How the candidate responded to the core screening questions and what competency signal those responses generated
• Overall recommendation: Pass / Conditional Pass / Decline, with the specific criteria outcomes that drove the recommendation
• Full call transcript: A complete text record of the conversation, available for human review
• Call summary narrative: A synthesized paragraph describing the candidate's screening performance in plain language

This output is pushed directly to the candidate's ATS profile, no recruiter data entry required.

One of BrewVoice's most significant design differentiators for the Indian market is its multilingual and multi-register capability.

India's professional talent pool is linguistically diverse. Candidates from different regions speak with different accents, switch between Hindi and English within a single sentence (Hinglish), use regional vocabulary, and apply different conversational conventions than candidates from other markets. A voice screening system built purely on US or UK English ASR and language models performs poorly with this candidate population, producing transcription errors, misunderstood responses, and evaluation inaccuracies that systematically disadvantage qualified candidates.

BrewVoice is trained and calibrated specifically for:

• Indian English in its regional variants (Delhi, Mumbai, Bengaluru, Chennai, Kolkata accent profiles)
• Hinglish code-switching, seamlessly handling candidates who move between Hindi and English within the same response
• Professional register variation, understanding when a candidate is using formal versus conversational language

This linguistic calibration is not a feature, it is a requirement for the system to function accurately with the candidate population it serves.

BrewVoice's interviewer persona is Dan, SkillBrew.AI's AI interviewer identity. Candidates interact with Dan throughout the screening call: it is Dan who introduces the call, asks the questions, follows up where needed, and closes the conversation. The Dan persona is professional, warm, and clearly AI-disclosed, candidates are never left uncertain about who or what they're speaking to.

BrewVoice integrates directly with the SkillBrew.AI hiring pipeline and can connect to external ATS platforms. The screening trigger, output delivery, and candidate status updates all happen within the workflow layer, recruiters see screening summaries appear in their queue without any manual data management between the call and the recruiter's review.

See BrewVoice in action →