Two modes. One goal — the right answer.
OpenInsight has two query modes, optimised for how doctors actually think: fast lookups and deep clinical reasoning.
Fast Search
The quick clinical colleague
Response time: ~3 seconds
Best for: DDI checks, dosing, differential lists, quick protocol lookups
Process: Embed query → hybrid retrieval → reranking → LLM synthesis
DeepInsight
The thorough consultant
Response time: 20–60 seconds
Best for: Complex cases, treatment planning, rare disease workup
Process: Multi-agent orchestration with RAG, web search, synthesis, and validation
How We Compare
Built for India. Different by design.
See how OpenInsight stacks up against generic AI tools and established Western references — across the capabilities Indian doctors actually need.
| Capability | Generic AI ToolsChatGPT, Gemini, etc. | UpToDate / WesternEstablished references | RecommendedOpenInsightBuilt for India |
|---|---|---|---|
| India-first evidence (ICMR / NTEP / NVBDCP) | Defaults to FDA / Western sourcesNo | Limited India coveragePartial | Ground-up ICMR / NTEP / NVBDCPYes |
| CDSCO drug database | FDA / EMA labels onlyNo | Lexicomp, not CDSCONo | Full CDSCO-approved genericsYes |
| Drug interactions (Indian generics) | Western brand namesNo | Generic name onlyPartial | Indian brand + Jan Aushadhi awareYes |
| Cost for Indian doctors | ₹2,000+ / month or USD-denominatedNote | ~₹18,000+ / year (institutional)Note | Free for verified NMC-registered doctorsYes |
| Offline capability | Online onlyNo | Limited cached contentPartial | On-device cache — pilot rolloutPartial |
| Citation export (PDF for patient records) | No clinical PDF exportNo | Print-friendly topicsPartial | Citation-ready PDF + VaultYes |
| Hindi language support | Auto-translate, lossyPartial | English-onlyNo | Native Hindi + 4 regional languages (roadmap)Yes |
| NMC / MCI verification & compliance | No practitioner verificationNo | Institutional licensingPartial | NMC registration verified at sign-upYes |
| DeepInsight multi-agent mode | Single-pass generationNo | Static topic articlesNo | RAG + Web + Synthesis + Validator agentsYes |
Full support Partial / on roadmap Not available
Behind the Scenes
How DeepInsight thinks
Receives the query, decides which agents to invoke and in what order. Manages escalation signals and coordinates the entire pipeline for optimal response quality.
Retrieves relevant chunks from the Zilliz vector store using hybrid dense+sparse search with evidence-level boosting to prioritize high-quality sources.
Queries live web for recent publications, drug approvals, and guideline updates not yet in the corpus for real-time information.
Receives outputs from RAG and Web Search agents, synthesises a coherent, structured clinical response ready for physician review.
Verifies every claim against its source. Flags unverified statements to maintain clinical credibility and safety.