What is Cekura?
Cekura (cekura.ai) is an automated QA platform for voice and chat conversational agents. Their product spans scenario-based testing, evaluations, and production observability (monitoring, analytics, and alerting on live traffic). These docs cover how to connect Vatel AI into that workflow. From an agent’s Testing tab in Vatel you copy chat credentials (WebSocket URL and session secret) or voice details (SIP trunk digest username and password, plus the assignment SIP URI). Cekura uses those to reach your agent, run tests, and show results in Cekura’s own UI. For scenario libraries, caller personas, parallel calling, LLM judges, and production observability, see what they publish on cekura.ai.How it helps
- Ship with confidence: Run repeatable checks after prompt, flow, or tool changes instead of relying only on manual spot calls.
- Exercise the agent “brain” over chat: Validate reasoning, branches, and tool calls without paying for a full voice session.
- Test the real voice path: Place SIP calls into Vatel so audio, telephony, and STT/TTS behave like a live caller (normal voice usage and billing).
Guides
Chat Testing
Text sessions over the Testing WebSocket and secret. Lighter footprint; ideal for logic, tools, and regressions without full voice cost.
SIP Testing
Inbound SIP into a generic trunk and agent assignment. End-to-end voice—the same path and cost profile as a production caller.