Automated Candidate Scoring via Voice AI: Post-Call Workflows

Bolti is a voice AI platform for building production-ready conversational phone agents that helps you automate outbound sales, customer support, and recruitment. For recruiting teams handling high-volume hiring in 2026, manually listening to phone screens is a massive bottleneck. By combining Bolti's automated HR screening features with a structured post-call webhook workflow, you can score and filter candidates automatically without picking up the phone. You can test this workflow yourself with Bolti's free trial which includes 50 minutes of calling time.

What is automated candidate scoring voice ai?

Automated candidate scoring via voice AI is the process of using structured call data, transcripts, and custom evaluation criteria to automatically grade a candidate's fit for a role. Instead of a recruiter spending 20 minutes on an initial phone screen, an AI agent conducts the interview, and a downstream system evaluates the transcript against your rubric.

This setup allows HR teams to instantly filter out unqualified applicants and fast-track top-tier talent to hiring managers. By utilizing Bolti's native HR screening module, you can easily turn resumes and job descriptions into structured voice calls.

How to structure the post-call evaluation payload

To automate candidate scoring, you need to capture the call data as soon as the conversation ends. Bolti provides real-time updates via webhooks, allowing you to process the call transcript immediately.

1. Subscribe to the completion event

Configure your system to listen for the conversation.completed event. This event fires as soon as a screening call transitions to a completed state and has been billed.

2. Capture the core transcript payload

When the webhook fires, Bolti sends a payload containing the metadata of the call and the full transcription. The structure typically includes:

• Candidate metadata: Name, email, and phone number parsed from the original CV.
• Call transcript: The turn-by-turn text of what the candidate and the AI agent said.
• Custom question answers: The specific responses to the questions you configured on the role.

3. Send the data to your LLM evaluation engine

Once your webhook receiver captures the payload, pass the full transcript and the target job description to an LLM (such as GPT-4o or Claude 3.5 Sonnet) with a strict scoring prompt. You can see how other companies structure their workflows in our case studies.

Step-by-step: Automating the scoring pipeline

Building an automated candidate scoring pipeline requires connecting Bolti's outbound calling capabilities to your internal ATS (Applicant Tracking System) or database. Here is how to set up the flow:

1. Create the role in Bolti: Go to Agents → HR Screening in your dashboard and click + New Role. Enter your job description. This serves as the baseline context for the AI agent.
2. Add custom questions: In the configuration panel, add questions like "What is your notice period?" or "What is your expected CTC?". These questions are injected directly into the agent's prompt via the {{ custom_questions }} template variable.
3. Upload candidate CVs: Upload your candidate pool. Bolti parses the resumes, generates refined bullet points, and populates the {{ candidate_details }} template variable.
4. Initiate the call: Schedule outbound calls or send self-booking links. Bolti will handle the call using low-latency STT and natural TTS voices, ensuring a professional candidate experience.
5. Process and score: When the call completes, your webhook endpoint receives the transcript. Run your evaluation prompt, calculate a score from 1 to 100, and push the score back to your ATS.

Designing the scoring prompt for your LLM evaluation

To get reliable, objective scores, your evaluation prompt must be highly structured. Avoid generic instructions like "Is this candidate good?" Instead, define specific scoring rubrics.

Use a system prompt similar to this:

"You are an expert technical recruiter. Review the attached transcript of a screening call for the Senior Backend Engineer role. Grade the candidate on a scale of 1 to 5 for each of the following categories: Technical Knowledge, Communication, and Alignment on Notice Period. Output your final response in a clean JSON format containing the category scores, a 2-sentence summary, and a boolean flag for 'pass_to_next_round'."

By enforcing a JSON output from your evaluation engine, you can easily parse the results programmatically. If the candidate receives a 'pass_to_next_round' flag, your system can automatically trigger an email inviting them to the next technical interview.

Set up your first automated screening agent

With Bolti, you can build a fully automated HR screening funnel in under 15 minutes. Our pay-as-you-go pricing starts at just ₹7/minute, making it incredibly cost-effective to scale your hiring operations. You can also start your free trial today to get 50 minutes of free call time and test your automated candidate scoring system end-to-end.

Frequently Asked Questions

What webhook event should I use to trigger candidate scoring?

You should subscribe to the conversation.completed event. This event fires as soon as the phone call ends and the transcription is finalized, providing you with the complete payload needed for scoring.

Does Bolti support Indian languages for candidate screening?

Yes. Bolti is built to be multilingual and supports Hindi, Marathi, Tamil, Telugu, Gujarati, Bengali, and English, along with over 80 global languages. For Indian languages, you can pair the agent with specialized STT providers like Fennec for high accuracy.

How do I pass candidate-specific information to the screening agent?

When you upload a CV, Bolti automatically parses the data. It then fills in template variables such as {{ candidate_name }} and {{ candidate_details }} in your agent's system prompt, ensuring the agent knows who it is speaking to and their background.

Can I connect Bolti to my existing ATS?

Yes. Every action in the Bolti dashboard is supported by our open REST API. You can use webhooks to send completed call data directly to your ATS or use our developer APIs to trigger calls automatically when a candidate reaches a specific stage in your pipeline.