Tatva Labs Inc. (d/b/a Cekura AI)'s sub-processors

Read automatically from Tatva Labs Inc. (d/b/a Cekura AI)'s own list on 28 October 2026. A reading can miss or misread a line: check the original before you rely on it.

Sub-processor	Purpose, as listed	Country
AWS	IT infrastructure	US
OpenAI	LLM inference (simulation, scenario generation, evaluation)	US
Microsoft Azure	Cloud computing service used for LLM inference - Azure OpenAI Service	US
Datadog	Monitoring and security platform for cloud applications	US
twillio.com	Telephony: Cekura-provided phone numbers placing/receiving test calls	US
daily	WebRTC rooms for testing Pipecat / WebRTC agents	US
Supabase	Open-source backend-as-a-service platform with real-time databases	US
Modal Labs	Serverless compute for sandboxed agent execution	US
Autoloops	—	US
Wafer	Custom DeepSeek model inference	US
Baseten	Hosted model inference - DeepSeek, Pi chunk-worker	US
Deep Infra	LLM inference - DeepSeek, Gemma	US
Fireworks AI	LLM inference - Kimi K3 (agent runtime)	US
Anthropic	LLM inference (Claude) for simulation and evaluation	US
Deepgram	AI-powered speech recognition platform for transcription services	US
ElevenLabs	Text-to-speech for simulated caller voices (voice processing)	US
Google Cloud	LLM inference - Gemini 2.5 Flash / Flash-Lite / Pro	US
Cartesia	Text-to-speech for simulated caller voices	US