Private by design
Checksum-validated PII redaction for the Americas, Europe, Asia-Pacific, India and Africa in any script, optional name redaction, pseudonymous authors, an encrypted database and air-gapped egress by default.
Voice of Customer without violating privacy
Collect feedback from every channel and language, remove personal data before it is stored, and let your AI agents answer questions with cited evidence. All inside your own infrastructure.
Why obsei
Section titled “Why obsei”Bring your own
App stores, helpdesks, calls, SQL warehouses, webhooks, communities and any REST or MCP source. Any OpenAI-compatible model, from Ollama to Azure OpenAI, or Bedrock and Vertex via LiteLLM.
AI-native
A read-only MCP server and a Claude plugin, so agents answer “what are customers saying?” with quotes and record ids, never author identities.
Every language
Feedback is classified and quoted in its own language; a local multilingual model groups the same issue across 50+ languages.
Start here
Section titled “Start here”QuickstartInstall, run the ten-language sample and ask your first question.
ConfigurationPipelines, scheduling, privacy and models in obsei.yaml.
Sources and sinksEvery built-in connector and how to map your own data.
Privacy controlsRedaction by region, pseudonyms, erasure, retention and the audit log.
MCP and agentsConnect Claude, ChatGPT, Cursor and your own agents.
Server and accessWebhooks, scheduling, Studio, SSO and roles with obsei serve.