Companies

Cekura

cekura.ai

Cekura tests and monitors voice and chat AI agents across the conversational AI lifecycle.

HQSan Francisco, California, United States
Employees1-50
1 active role
Jobs checked 22h ago
ObservabilityAI InfrastructureB2B SaaS

About

Cekura builds a testing and observability platform for conversational AI, covering voice and chat agents from pre-production simulation and evaluation through production call monitoring and CI/CD support. It sells primarily to conversational-AI companies and enterprises across healthcare, BFSI, logistics, recruitment, and retail, differentiating through automated scenario generation, lifecycle monitoring, and actionable quality metrics.

Market

Cekura competes in the conversational-AI quality, testing, evaluation, and observability market, with a particular emphasis on production voice AI. It positions itself as an end-to-end reliability layer spanning pre-production simulation, voice-specific infrastructure and conversation tracing, production-call QA, security testing, and continuous evaluation; its differentiation is the combination of voice-aware signals across audio, STT, LLM reasoning, tool calls, and TTS with native integrations across major voice-agent stacks.

Target Customers

Cekura targets growth-stage and enterprise companies deploying customer-facing voice and chat AI agents, particularly in healthcare, BFSI, logistics, recruitment, retail, fintech, and contact-center environments. Its primary users and buyers are engineering, QA, product, and operations teams responsible for agent reliability, compliance, deployment velocity, and production-call quality.

At a Glance

Problem

Conversational AI agents can fail in ways that conventional software testing or basic call review may miss, including gibberish, interruptions, latency, poor sentiment, and other voice-specific quality issues. These failures are especially costly in customer-service and outbound calling, where a bad interaction can damage customer experience and undermine confidence in automation. The core pain is maintaining reliable performance at high call volumes: teams need to find regressions before launch and identify problems across live conversations, rather than discovering them one customer at a time. The research does not publish a specific dollar cost or ROI, but the killer use case is clear—stress-test a voice agent against difficult customer personas before deployment, then continuously detect failures in production.

Product / Service

Cekura is an end-to-end testing and observability platform for conversational AI. Before launch, teams run simulated conversations across diverse personas; after launch, Cekura monitors real calls and evaluates them using voice-specific signals such as gibberish detection, interruption tracking, latency, sentiment, and pitch. Its workflow also includes stereo recordings, real-time metrics, customizable LLM judges that can be tuned against ground truth, production dashboards, conversation analytics, and alerts through Slack, email, or webhooks.

The benefit is a continuous reliability loop rather than a one-time QA exercise: teams can simulate, inspect, score, monitor, and improve their agents from one system. Cekura is offered as a software platform with a free-trial entry point and a demo-led sales option, and it integrates with major voice AI stacks including LiveKit, Pipecat, Vapi, Retell, and ElevenLabs.

Market

Cekura competes in the emerging conversational-AI reliability market, specifically AI voice-agent testing, evaluation, and production observability. The available research does not identify named direct competitors. Instead, Cekura positions itself as a cross-stack reliability layer that integrates with the major voice-agent platforms, which are better understood from the evidence as ecosystem partners or adjacent platforms than as confirmed competitors.

Cekura is not presented as pre-revenue. Its June 2025 fundraising announcement reported a $2.4 million seed round, Y Combinator backing, more than 60,000 voice-AI calls evaluated daily, and more than five million voice-agent minutes stress-tested; its site also claims a Product Hunt #1 launch. A third-party directory estimates approximately $770,000 in annual revenue, though that figure is not audited in the available research. The company was listed at roughly 15 people, indicating an early but commercially active startup rather than a mature incumbent.

Founders & Leadership

Tarush AgarwalFounder
CEO
Shashij GuptaFounder
CTO
Sidhant KabraFounder
CBO

Funding History

2025-06
Seed$2.4M

Y Combinator

Recent News

2026-06-04partnership
Native ElevenLabs Conversational AI Integration

Cekura announced native integration with ElevenLabs Conversational AI, enabling automated scenarios, regression testing, and continuous production observability for ElevenLabs agents.

2026-05-25
The 11 Best Voice Agent Testing Platforms in 2026

Speechmatics’ industry roundup positions Cekura, formerly Vocera, as a broad QA platform for voice and chat agents, particularly suited to structured simulation and CI integration.

2026-05-24product
May Week 4 Product Updates: Optimize Agent and EU Deployment

Cekura launched an Optimize Agent feature that uses evaluators to suggest targeted prompt improvements. The company also announced EU deployment for lower latency and data-residency support, alongside PDF exports and additional integration improvements.

2026-03-31partnership
Speechmatics and Cekura Bring Real-World STT Testing to Voice Agent Pipelines

Speechmatics and Cekura announced an integration embedding Speechmatics’ speech-to-text engine into Cekura’s testing and production-monitoring platform. The integration supports testing noisy audio, dialects, multi-speaker conversations, and comparisons with other STT providers.

2026-03-16product
March Week 3 Product Updates: LiveKit Tracing Integration and Retell WebRTC Testing

Cekura added LiveKit tracing through its Python SDK, capturing transcripts and tool calls for evaluation. The release also included Retell WebRTC testing and alert-routing improvements.

2026-01product
January Week 5 Product Updates: VAPI, Retell, ElevenLabs, and Slack Improvements

Cekura improved its VAPI, Retell, and ElevenLabs integrations with automatic production-call fetching and metadata collection, and added a Slack integration.

2025-11product
November Week 4 Product Updates: Agentforce and Kore AI Integrations

Cekura added support for Agentforce chatbot integration and Kore AI voice agents, expanding the platforms that can be tested through its QA tooling.

2025-11-10product
November 10 Product Updates: Pipecat WebRTC and ElevenLabs WebSocket Integrations

Cekura added a UI-driven Pipecat WebRTC integration and support for connecting to ElevenLabs agents over WebSockets for voice conversations instead of telephony.

2025-10product
October Week 3 Product Updates: No-Code LiveKit Integration

Cekura introduced no-code LiveKit integration for testing, automatically creating required rooms and connecting to agents over WebRTC. The release also highlighted expanded tool-call testing.

2025-09product
September Week 1 Product Updates: Bland Integration

Cekura added a native Bland integration for voice-agent testing and expanded knowledge-base support to formats including JSON and CSV.

Active Roles

1
San Francisco, CA, US/Sales/Today

Business Model

Cekura monetizes through a month-to-month SaaS subscription. Its pricing page advertises a $500/month plan, 300 free credits, and $30/month per additional seat.

Products

Pre-production voice and chat AI testing with synthetic scenarios, diverse personas, and workflow simulationsProduction voice observability and call-quality monitoringAutomated QA, custom metrics, LLM judges, and continuous evaluationSecurity testing and red teaming for conversational AI agentsMCP server and developer tooling for recurring tests, deployment-triggered evaluations, and CI/CD workflows

Customers

Twin HealthLindyQuo

Tech Stack

LLM-based evaluation and judgingSpeech-to-text (STT) and text-to-speech (TTS) pipelinesVoice infrastructure and WebRTCModel Context Protocol (MCP)API and CI/CD integrations for automated testing

Competitors

Hamming
Coval
Roark
Coglity
Privo
Braintrust
Botium