Automate Candidate Screening and Scheduling for Consultancies

Most recruiting teams rely on their ATS, like Greenhouse or Lever, for initial filtering. These tools use simple keyword matching, which cannot understand context. A resume with "Java" is flagged for a "JavaScript" role, creating noise and requiring manual review for every single applicant.

A 25-person tech company posts a "Senior Python Developer" role and gets 200 applicants. Their recruiter uses Lever's filters for "Python" and "API" but still has 80 resumes to read by hand. After a full day of reading, they email a Calendly link to 20 candidates. 15 book a call, but 5 of them are clearly not senior level, wasting 2.5 hours on screening calls that should have been an email.

Off-the-shelf chatbots from vendors like Paradox are designed for enterprises with massive volume. They are expensive, have rigid conversation flows you cannot customize for niche roles, and charge high monthly fees based on candidate count. For a growing company, this approach is too costly and inflexible.

Syntora would approach the problem of automating initial candidate screening and scheduling by first conducting a discovery phase to understand your specific ATS setup, integration requirements, and unique hiring workflows. We would then design and implement a custom technical architecture tailored to your needs.

The core of the system would involve integrating with your ATS API, such as Lever or Greenhouse, to pull new candidate data in real-time. We would leverage Python with the httpx library for robust API communication. Each candidate's resume would be parsed to extract key information, then structured and stored in a Supabase Postgres database. This structured data forms the foundation for efficient processing and retrieval.

Next, a "job fingerprint" would be created for each role. This involves using the Claude API to analyze your job descriptions, extracting key skills, required experience levels, and essential qualifications. For each candidate, a similar fingerprint would be generated from their parsed resume using a Python script, often deployed on an AWS Lambda function for scalable processing. We would then calculate a similarity score between the job and candidate fingerprints. Candidates exceeding a predefined threshold would proceed to the interview stage. We've built document processing pipelines using Claude API for financial documents, and the same pattern applies to extracting insights from resumes and job descriptions.

For qualified candidates, a FastAPI service would power a personalized email invitation to a text-based interview. The interview experience would be designed using the Claude API, incorporating 5-7 specific screening questions tailored to the role, focusing on project experience and technical depth. All conversation transcripts would be securely logged back to Supabase for auditability and further analysis.

Upon successful completion of the interview, the system would utilize the Google Calendar API to identify and present available interview slots from the hiring manager's calendar. Once a time is confirmed by the candidate, a calendar invite would be automatically created. Finally, a summary of the AI interview would be pushed as a detailed note to the candidate's profile within your ATS, ensuring a seamless handover to your human recruiters. The goal for this automated flow would be to significantly reduce the manual effort involved in candidate screening and scheduling.