One search platform, every surface.
From a marketing site's command-palette to a clinic's chart-search bar, the same primitives do the work: Lunexa-grade relevance, Smart Crawler, NL Search, scoped API keys, and an architecture designed for HIPAA workloads. Every use case below is grounded in features that ship today — not roadmap hand-waves.
Site search
Drop a fast, typo-tolerant search bar into any website. One SDK, one API key, real results in under an hour.
Documentation & help-center search
Point Smart Crawler at your docs. It builds the schema, ingests the pages, and keeps them fresh.
E-commerce product discovery
Faceted filters, synonyms, and curations built for catalogs that change daily.
Headless commerce
API-first by design. Your storefront stays yours — Next, Remix, Astro, Hydrogen, native apps, all equal citizens.
SaaS in-app search
Command-palettes, global search, tenant-scoped lookups — the backbone for every modern SaaS UI.
Media & content discovery
Blogs, newsrooms, video catalogs. Facet by topic, sort by recency, let readers ask in natural language.
Natural-language search
Let users ask full questions. Lunexa extracts filters and intent, returns ranked results — not a chat window full of hallucinations.
Internal & enterprise knowledge
Point Smart Crawler at your public wiki, help center, or intranet. Search across them from one bar.
Healthcare records & clinical search
Search patient charts, clinical notes, lab results, and diagnostic codes — fast — with an architecture designed for HIPAA workloads and a BAA available on Enterprise.
Patient & member portals
Member-facing search over appointments, claims, prescriptions, and provider directories — with PHI exposure scoped to the logged-in member only.
Location & geo-search
Sort and filter by distance. Storefronts, clinics, pharmacies, dealerships, ATMs — anything pinned to a coordinate.
Recommendations & similarity
Build "people also viewed," related-content, and personalized rails on top of vector embeddings.
Bring your own LLM via MCPNew
Connect Claude Desktop, ChatGPT, Cursor, or Continue.dev to your Lunexa project with one API key. Your LLM searches your data through the Model Context Protocol — no scraping, no sync, no copy.
Customer-support copilotNew
Drop a Lunexa-MCP integration into Claude Desktop or ChatGPT and your support team gets a copilot that searches the actual knowledge base, ticket history, and product docs — instead of hallucinating from training data.
Developer assistant in your IDENew
Cursor and Continue.dev pick up Lunexa as an MCP server with one config line. Your engineers ask 'how do I call the X API?' inside the editor and the LLM grounds its answer in your actual API docs, code samples, and ADRs.
Research / RAG agent grounded in your dataNew
Build a research agent in your own LLM workspace (Claude Projects, ChatGPT custom GPT, internal Bedrock app) that runs a multi-step plan against Lunexa: search → narrow → fetch → synthesize. No bespoke RAG plumbing.
Site search
Drop a fast, typo-tolerant search bar into any website. One SDK, one API key, real results in under an hour.
- Sub-50 ms queries on the Lunexa search infrastructure.
- Typo tolerance, synonyms, and stop-words configurable per collection.
- Works with any framework via the @lunesearch/client SDK.
Documentation & help-center search
Point Smart Crawler at your docs. It builds the schema, ingests the pages, and keeps them fresh.
- Smart Crawler auto-detects page patterns and field types.
- Scheduled re-crawls keep the index in step with your docs site.
- NL Search lets readers ask full questions instead of guessing keywords.
E-commerce product discovery
Faceted filters, synonyms, and curations built for catalogs that change daily.
- Facet on any field: price, brand, size, in-stock, tags.
- Curations promote seasonal or sponsored items without rewriting the ranking.
- Per-collection cache TTL cuts repeat-query latency and infra cost.
Headless commerce
API-first by design. Your storefront stays yours — Next, Remix, Astro, Hydrogen, native apps, all equal citizens.
- REST + SDK, no lock-in to a storefront framework.
- Scoped API keys per channel (web, mobile, PoS) with per-key rate limits.
- Nonprod / production environment tags keep staging traffic off your live index.
SaaS in-app search
Command-palettes, global search, tenant-scoped lookups — the backbone for every modern SaaS UI.
- Per-tenant API keys with search filters baked in — no leakage between customers.
- Multi-project workspace keeps dev / staging / prod cleanly separated.
- Real-time analytics surface zero-result queries you need to fix.
Media & content discovery
Blogs, newsrooms, video catalogs. Facet by topic, sort by recency, let readers ask in natural language.
- Multi-field search across title, body, tags, author, transcript.
- NL Search parses intent like "interviews from last month about climate" into filters.
- Smart Crawler keeps the index current without a publishing-side integration.
Natural-language search
Let users ask full questions. Lunexa extracts filters and intent, returns ranked results — not a chat window full of hallucinations.
- LLM-backed query understanding; keyword fallback when confidence is low.
- Separate NL cache TTL (default 24 h) keeps embedding cost bounded.
- Grounded in your collection — no free-form generation, no invented facts.
Internal & enterprise knowledge
Point Smart Crawler at your public wiki, help center, or intranet. Search across them from one bar.
- One collection per source or one unified collection with a source facet — your call.
- Super-admin controls + audit trail for compliance-sensitive teams.
- Per-project isolation so Engineering's index never surfaces in Sales' UI.
Sales-led on Enterprise
BAA, security questionnaires, audit-log retention contracts, and custom SLAs are part of the conversation. Reach out and we'll line up a security review with the right docs for your compliance team.
Healthcare records & clinical search
Search patient charts, clinical notes, lab results, and diagnostic codes — fast — with an architecture designed for HIPAA workloads and a BAA available on Enterprise.
- HIPAA-aligned architecture: encryption in transit (TLS 1.2+) and at rest, single-tenant project isolation, scoped API keys, full audit trail.
- Field-level access via scoped keys — billing staff see CPT and ICD-10, clinicians see notes, nobody sees more than their role allows.
- Sub-50 ms lookups across millions of encounters keep the chart open and the clinician in flow.
- Synonyms map clinical shorthand to formal terminology ("MI" → myocardial infarction, "HTN" → hypertension) so search works the way clinicians actually type.
- Business Associate Agreement available on Enterprise (in roll-out). Audit log retained per your retention policy; export to your SIEM.
Sales-led on Enterprise
BAA, security questionnaires, audit-log retention contracts, and custom SLAs are part of the conversation. Reach out and we'll line up a security review with the right docs for your compliance team.
Patient & member portals
Member-facing search over appointments, claims, prescriptions, and provider directories — with PHI exposure scoped to the logged-in member only.
- Per-member scoped API keys carry a baked-in filter (member_id = X) — a leaked key can only ever return that member's records.
- Provider directory search with geo-radius + facets (specialty, accepted plans, languages, telehealth-only) for in-network lookups.
- Natural-language queries handle the way real members ask ("physical therapist near me that takes my plan") without breaking on synonyms or word order.
- Production environment label gates production collections from accidental deletion; non-prod traffic stays off the live PHI index.
Sales-led on Enterprise
BAA, security questionnaires, audit-log retention contracts, and custom SLAs are part of the conversation. Reach out and we'll line up a security review with the right docs for your compliance team.
Location & geo-search
Sort and filter by distance. Storefronts, clinics, pharmacies, dealerships, ATMs — anything pinned to a coordinate.
- Native geopoint type plus geo_distance sort and bounding-box filters.
- Combine geo with regular facets — "open now, accepts insurance, within 5 miles" is a single query.
- Per-collection cache keeps repeat "near me" queries off the engine on busy days.
Recommendations & similarity
Build "people also viewed," related-content, and personalized rails on top of vector embeddings.
- Vector fields hold embeddings from any provider — OpenAI, Cohere, Voyage, your own model.
- Hybrid scoring blends keyword relevance with semantic similarity in a single ranked list.
- NL Search reuses the same embedding column for ask-style questions — one source of truth, two surfaces.
Bring your own LLM via MCPNew
Connect Claude Desktop, ChatGPT, Cursor, or Continue.dev to your Lunexa project with one API key. Your LLM searches your data through the Model Context Protocol — no scraping, no sync, no copy.
- Read-only tool surface: keyword search, multi-search, document fetch, schema introspection — scoped to your project by API key.
- Auth via existing Lunexa API keys with project + collection scopes — leaks stay contained the same way as the REST API.
- 20,000 included tool calls per month on Enterprise; metered overage at $2.00 / 1,000 with a 2× quota hard cap as runaway protection.
- Same compliance posture as the rest of the platform — encryption in flight + at rest, audit log of every call, BAA on Enterprise.
Sales-led on Enterprise
BAA, security questionnaires, audit-log retention contracts, and custom SLAs are part of the conversation. Reach out and we'll line up a security review with the right docs for your compliance team.
Customer-support copilotNew
Drop a Lunexa-MCP integration into Claude Desktop or ChatGPT and your support team gets a copilot that searches the actual knowledge base, ticket history, and product docs — instead of hallucinating from training data.
- Agent calls lunexa_search across your KB collection, then lunexa_get_document on the specific article — every reply has citations.
- Multi-search runs the same query across docs + past tickets + product changelog in one round-trip; the LLM picks the best source.
- Per-org API keys mean the copilot only ever sees the tenant's own collections — no cross-customer leakage.
- Every tool call logs to the admin audit feed, so QA can replay exactly what context the agent used to draft a reply.
Sales-led on Enterprise
BAA, security questionnaires, audit-log retention contracts, and custom SLAs are part of the conversation. Reach out and we'll line up a security review with the right docs for your compliance team.
Developer assistant in your IDENew
Cursor and Continue.dev pick up Lunexa as an MCP server with one config line. Your engineers ask 'how do I call the X API?' inside the editor and the LLM grounds its answer in your actual API docs, code samples, and ADRs.
- Index API reference, internal ADRs, and runbooks into one Lunexa project; expose a read-only API key to your team's IDE config.
- lunexa_list_collections lets the LLM plan: 'should I search adrs or api-docs first?' — no hardcoded routing in the prompt.
- Field-scoped search keeps internal-only docs out of the response when the developer hits a customer-facing collection by mistake.
- Local IDE → MCP over HTTPS; nothing in the prompt cache leaves the customer's vendor (Claude / OpenAI) stack.
Sales-led on Enterprise
BAA, security questionnaires, audit-log retention contracts, and custom SLAs are part of the conversation. Reach out and we'll line up a security review with the right docs for your compliance team.
Research / RAG agent grounded in your dataNew
Build a research agent in your own LLM workspace (Claude Projects, ChatGPT custom GPT, internal Bedrock app) that runs a multi-step plan against Lunexa: search → narrow → fetch → synthesize. No bespoke RAG plumbing.
- Lunexa replaces the 'retriever' half of a RAG stack: facets, filters, sort_by, and pagination all available to the LLM as tool args.
- Multi-search ships up to 8 parallel queries in one MCP call — perfect for an agent that wants to triangulate before answering.
- Per-call $0.002 list price + 20,000 included on Enterprise — predictable cost vs. spinning up your own retrieval service.
- Hard cap at 2× quota stops a runaway tool-loop from billing-bombing you while you debug a prompt.
Sales-led on Enterprise
BAA, security questionnaires, audit-log retention contracts, and custom SLAs are part of the conversation. Reach out and we'll line up a security review with the right docs for your compliance team.
Built for regulated workloads.
Encryption in transit (TLS 1.2+) and at rest, single-tenant project isolation, scoped API keys with bake-in filters, and a full audit trail of admin + data-plane events. HIPAA programme is in active roll-out on Enterprise — Business Associate Agreements available to qualifying customers as that programme reaches GA. SOC 2 Type II is on the same track.
Need the BAA today, or a security questionnaire? Talk to us — Enterprise customers get prioritized review.
Not on this list? Good question.
Image search, reverse visual lookup, and sponsored-listing ad networks are intentionally absent. They're real use cases — just not ones Lunexa does well today. We'd rather skip them than sell you a placeholder. If you need them, tell us — that's how we pick what to build next.
Pick a use case. Ship it this week.
Every plan includes the full platform. Start on the free tier and only upgrade when your traffic says so.