Get Started
Teamcast Platform
AI-led async behavioral interviews with defensible, coverage-only measurement — built for recruiting teams and ATS/HRIS partners.
Teamcast runs the interview end-to-end: a recruiter (or a partner system, via the API) submits a role and a candidate, Teamcast's AI interviewer Maya conducts a live, adaptive voice interview, and the platform returns a structured, deterministic assessment — with a human able to approve the plan and the assessment at each gate.
For integration partners, Teamcast exposes the same operations two ways: a REST Integration API under /integration/* for conventional HTTP clients, and a JSON-RPC 2.0 A2A surface at POST /a2a/task for agent-to-agent tooling. Both are backed by the same service, so a run created through one can be queried or approved through the other.
coverage_summary (sufficient / partial / insufficient plus gap_competencies) — never a numeric score, a hire/no-hire recommendation, or any inferred emotion signal. This is a hard platform constraint (EU AI Act aligned), not a per-tenant setting. See Privacy & Data.| Layer | Responsibility |
|---|---|
| Auth & Access | X-API-Key integration keys, OAuth 2.1 (client_credentials / authorization_code), user JWT — resolved to one unified principal + RBAC permission set |
| A2A / Integration API | Interview lifecycle: create, status, info-completion, plan approval, assessment approval, cancellation, GDPR purge |
| HITL Workflow | Two approval gates — plan generation and assessment — each rejectable/revisable before the run advances |
| Webhooks | HMAC-signed, retried delivery of interview lifecycle events to your registered callback |
What you can build
- ATS/HRIS integration — create interview runs from your applicant pipeline via REST or JSON-RPC, using a role + skills instead of a fixed skills taxonomy
- Webhook-driven pipelines — react to plan-generated, approved, and assessment-ready events instead of polling
- Sign in with Teamcast — OAuth 2.1 SSO so your product can act on a recruiter's behalf without sharing passwords
- PII-minimal workflows — use
candidate_refinstead of name/email, and purge candidate data on demand once the assessment is delivered