Everything you need
to deploy AI knowledge
Clara handles the infrastructure — embeddings, vector search, streaming, voice, analytics. You focus on uploading your knowledge.
Core capabilities
Everything ships out of the box. Nothing to build from scratch.
RAG-Powered Answers
Every response is grounded in your actual documents — no hallucinations. Clara cites exactly which paragraph it's drawing from, with source references like [S1] and [S2].
- Vector semantic search
- Multi-hop reasoning
- Source citations on every answer
- Hybrid BM25 + dense retrieval
Real-Time Voice AI
Full voice conversations powered by OpenAI Realtime API. Your customers speak naturally — Clara understands context, interruptions, and responds with low latency.
- WebRTC audio streaming
- 40+ languages supported
- Interrupt detection
- Custom voice personas
Any Document Type
Upload PDFs, Word docs, Excel spreadsheets, or paste a URL. Clara extracts, chunks, and indexes everything automatically using an advanced ingestion pipeline.
- PDF, DOCX, TXT, CSV
- URL / web scraping
- Automatic re-indexing
- 50MB per file limit
Multilingual
Clara understands queries in 50+ languages and responds in the language your customer prefers — even if your source documents are all in English.
- Auto language detection
- Cross-lingual retrieval
- 50+ supported languages
- Consistent citation quality
Fully Branded
Customise the assistant's name, avatar, colours, and personality to exactly match your brand. Embed as an iFrame or floating widget on any website.
- Custom assistant name
- Colour + logo theming
- System prompt control
- Tone configuration
Deep Analytics
See exactly what your customers are asking, how fast Clara responds, and where your knowledge base has gaps. Turn every query into a product insight.
- 30-day query charts
- Top query analysis
- Latency tracking
- Per-KB breakdown
API Access & Embed
Generate API keys with per-origin restrictions and rate limits. Embed Clara on any website with a single iFrame snippet or JavaScript loader.
- API key management
- CORS origin allowlisting
- Rate limiting per key
- Embed snippet generator
Multi-Tenant Architecture
One Clara deployment serves unlimited organisations, each with completely isolated data, their own knowledge bases, users, branding, and subscription tier.
- Full data isolation
- Per-org Qdrant namespace
- Role-based access control
- Org-level audit logs
Enterprise Security
Production-grade security built in from day one. JWT authentication, bcrypt password hashing, parameterised queries, MIME validation, and CORS protection.
- JWT + httpOnly cookies
- bcrypt password hashing
- SQL injection prevention
- File type validation
Streaming Responses
Chat responses stream token-by-token via NDJSON so users see text appearing immediately — not waiting for the full answer to generate.
- NDJSON token streaming
- < 300ms first token
- Graceful error recovery
- Connection keep-alive
SSO / SAML
Enterprise customers can integrate their existing identity providers — Azure AD, Okta, Google Workspace — for seamless single sign-on.
- SAML 2.0 support
- Azure AD / Okta
- Google Workspace
- Just-in-time provisioning
Automatic Re-Indexing
When you update a document, Clara automatically detects the change, re-processes the updated content, and updates the vector index without any manual steps.
- Change detection
- Background processing
- Zero downtime updates
- Processing status tracking
Feature comparison
See exactly what each plan includes