Open source · Self-hosted · Apache-2.0

Voice of Customer without violating privacy

Collect feedback from app stores, tickets, calls, surveys and communities, in every language. obsei removes personal data before anything is stored, finds the themes that matter, and lets your AI agents answer questions with cited evidence. All inside your own infrastructure.

$ uv tool install --prerelease allow "obsei[mcp]"
$ obsei init && obsei run
obsei Studio showing themes, a knowledge graph of sources, languages and intents, and feedback in eight languages
Studio with synthetic feedback in eight languages.

How it works

  1. CollectApp Store, Google Play, Zendesk, Intercom, Gong, GitHub, Bluesky, SQL, webhooks, files, any REST or MCP source.
  2. ProtectEmails, phones, cards, national IDs and optionally names are redacted at ingest; authors become pseudonyms.
  3. UnderstandYour model classifies sentiment and intent; stable themes and near-duplicates emerge across languages.
  4. ActSend to Slack, Jira, Linear, GitHub, webhooks or your warehouse. Only new or changed feedback.
  5. AskAgents query through MCP, people use Studio or Slack, and every answer cites its records.

Built for teams that cannot send customer data to someone else's cloud

Private by design

Checksum-validated PII redaction for the Americas, Europe, Asia-Pacific, India and Africa in any script. Encrypted database. Air-gapped by default.

Bring your own

Your sources, your keys, your model: Ollama, vLLM, Azure OpenAI, OpenAI, or Bedrock and Vertex through LiteLLM. Nothing leaves without your say-so.

AI-native

A read-only MCP server and a Claude plugin: agents answer "what are customers saying?" with quotes, record ids and no author identities.

Every language

Feedback is classified and quoted in its own language. A local multilingual model groups the same issue across 50+ languages.

Themes you can trust

Stable theme ids with trends, near-duplicate detection and k-anonymous views, so small groups never point at one customer.

Enterprise ready

Role-based access, SSO-proxy support, an audit log for erasure, export and evidence access, scheduled pipelines and a signed container image.

Controls that support privacy law

Access and erasure requests, retention, purpose tags and pseudonymisation map to rights in the GDPR and UK GDPR, India's DPDP Act, Brazil's LGPD, California's CCPA, Japan's APPI, South Africa's POPIA, China's PIPL and Singapore's PDPA. obsei makes no compliance claims: you remain the data controller.

Try it in five minutes

Ten languages of sample feedback, no API keys, nothing leaves your machine.