Homepage Messaging Update: "Turn Data Into Intelligence"

Turn Data Into Intelligence

Seven products that connect your conversations, documents, and knowledge—
so AI delivers insights, not just answers.

Your data is everywhere—conversations in Slack, documents in Drive, code in GitHub, 
tickets in Linear. AI can't connect the dots because each source is a silo.

**Papr transforms scattered data into connected intelligence** through seven products 
that work standalone or together:

- 📊 **Vector Memory** - Semantic search over any content
- 💬 **Chat Memory** - Conversation storage and compression
- 📄 **Document Intelligence** - Extract structure from PDFs and docs
- 🔗 **Knowledge Graphs** - Map relationships, not just similarity
- 🎯 **Graph-Aware Search** - Domain-tuned retrieval (code, science, etc.)
- 🤖 **AI Model Proxy** - Unified multi-model API
- 🔄 **Sync & Portability** - Local/cloud memory sync

[Explore products →](./products.md) | [Quick start →](../quickstart/index.md)

<Section id="products">
  <SectionTitle>Seven Products, One Platform</SectionTitle>
  <SectionSubtitle>
    Use standalone or combine them. Start simple, add intelligence as you need it.
  </SectionSubtitle>
  
  <Cards columns={3}>
    <Card
      title="Vector Memory"
      badge="Core"
      description="Semantic search and RAG foundation. Store memories, retrieve with natural language."
      link="/overview/products#vector-memory"
    />
    <Card
      title="Chat Memory"
      badge="Core"
      description="Conversation storage with automatic compression. Session management built-in."
      link="/overview/products#chat-memory"
    />
    <Card
      title="Document Intelligence"
      badge="Core"
      description="Extract structured information from PDFs, contracts, research papers."
      link="/overview/products#document-intelligence"
    />
    <Card
      title="Knowledge Graphs"
      badge="Addon"
      description="Map relationships between entities. Query with GraphQL. Build domain ontologies."
      link="/overview/products#knowledge-graphs"
    />
    <Card
      title="Graph-Aware Search"
      badge="Addon"
      description="Domain-tuned embeddings. Filter by language, topic, evidence type. +36% accuracy for science."
      link="/overview/products#graph-aware-search"
    />
    <Card
      title="AI Model Proxy"
      badge="Core"
      description="Call OpenAI, Anthropic, Google through one API. Track costs across providers."
      link="/overview/products#ai-model-proxy"
    />
  </Cards>
  
  <CallToAction>
    <Button href="/overview/products">Compare all products →</Button>
  </CallToAction>
</Section>

<Section id="outcomes" backgroundColor="light">
  <SectionTitle>From Fragments to Intelligence</SectionTitle>
  
  <Grid columns={2}>
    <Feature
      icon="🔍"
      title="Search that understands"
      description="Not just keywords—semantic search across conversations, docs, and code. 91%+ accuracy on Stanford's STaRK benchmark."
    />
    <Feature
      icon="🔗"
      title="Connected context"
      description="Code → ticket → conversation → decision. Your knowledge becomes one connected story, not isolated pieces."
    />
    <Feature
      icon="⚡"
      title="Predictive intelligence"
      description="Anticipates what users need before they ask. Pre-caches context for <150ms retrieval (when cached)."
    />
    <Feature
      icon="🎯"
      title="Domain-aware precision"
      description="Filter by programming language, scientific field, or custom dimensions. Search returns what you actually need."
    />
    <Feature
      icon="📊"
      title="Analytics and insights"
      description="Query relationships with GraphQL. Find patterns across your data. Answer 'why' not just 'what'."
    />
    <Feature
      icon="🔒"
      title="Enterprise-ready security"
      description="Built-in ACLs, namespace isolation, compliance controls. Data never leaks across users."
    />
  </Grid>
</Section>

Three ways to send data → One intelligence layer → Two ways to query

**Input:**
1. Documents (PDFs, Word docs)
2. Conversations (chat messages, sessions)
3. Direct memories (explicit data)

**Intelligence Layer:**
- Vector embeddings (semantic understanding)
- Knowledge graphs (relationship mapping)
- Predictive caching (anticipate needs)
- Domain tuning (specialized search)

**Output:**
1. Natural language search (ask questions, get insights)
2. GraphQL analytics (structured queries, aggregations)

[See architecture →](../concepts/architecture.md)

<Section id="proof">
  <SectionTitle>Trusted by AI Teams</SectionTitle>
  
  <Stats>
    <Stat
      number="#1"
      label="Stanford STaRK Benchmark"
      description="91%+ retrieval accuracy"
    />
    <Stat
      number="<150ms"
      label="Response time"
      description="With predictive caching"
    />
    <Stat
      number="7"
      label="Products"
      description="Standalone or combined"
    />
  </Stats>
  
  <Testimonial
    quote="Papr turned our scattered documentation into an intelligent knowledge base. 
           Our support team now surfaces answers 3x faster."
    author="Engineering Lead"
    company="Enterprise SaaS"
  />
</Section>

## Start Here

### 1. Choose Your Path

**Building something specific?**
- [Document Q&A](../tutorials/document-qa.md) - Extract + search PDFs
- [Conversational AI](../tutorials/chat-history.md) - Chat with memory
- [Code Search](../guides/graph-aware-embeddings.md) - Find code by intent
- [Knowledge Management](../tutorials/enterprise-saas.md) - Multi-tenant intelligence

**Just exploring?**
- [Products Overview](./products.md) - See all seven products
- [Quick Start (5 min)](../quickstart/index.md) - Ship a prototype
- [Decision Tree](./decision-tree.md) - Which products do you need?

### 2. Integrate

- [TypeScript SDK](../sdks/typescript.md)
- [Python SDK](../sdks/python.md)
- [REST API Reference](../apis/index.yaml)
- [LangChain / LlamaIndex](../examples/index.md)

# Turn Data Into Intelligence

**Seven products that connect your conversations, documents, and knowledge—so AI delivers insights, not just answers.**

> **TL;DR:** Store with `POST /v1/memory`. Search with `POST /v1/memory/search`.  
> Ranked **[#1 on Stanford's STaRK benchmark](https://huggingface.co/spaces/snap-stanford/stark-leaderboard)** 
> with **91%+ accuracy** and **<150ms retrieval** (when cached).

---

## From Data to Intelligence

Your data is everywhere—conversations in Slack, documents in Drive, code in GitHub, tickets in Linear. AI can't connect the dots because each source is a silo.

**Papr transforms scattered data into connected intelligence** through seven products that work standalone or together:

- 📊 **Vector Memory** - Semantic search over any content
- 💬 **Chat Memory** - Conversation storage and compression
- 📄 **Document Intelligence** - Extract structure from PDFs and docs
- 🔗 **Knowledge Graphs** - Map relationships, not just similarity
- 🎯 **Graph-Aware Search** - Domain-tuned retrieval (code, science, etc.)
- 🤖 **AI Model Proxy** - Unified multi-model API
- 🔄 **Sync & Portability** - Local/cloud memory sync

[Explore all products →](./products.md) | [Quick start (5 min) →](../quickstart/index.md)

---

## Intelligence Outcomes

### Search that understands
Not just keywords—semantic search across conversations, docs, and code. **91%+ accuracy** on Stanford's STaRK benchmark.

### Connected context
Code → ticket → conversation → decision. Your knowledge becomes **one connected story**, not isolated pieces.

### Predictive intelligence
Anticipates what users need before they ask. Pre-caches context for **<150ms retrieval** (when cached).

### Domain-aware precision
Filter by programming language, scientific field, or custom dimensions. Search returns **what you actually need**.

### Analytics and insights
Query relationships with GraphQL. Find patterns across your data. Answer **"why"** not just **"what"**.

### Enterprise-ready security
Built-in ACLs, namespace isolation, compliance controls. **Data never leaks** across users.

---

## Seven Products, One Platform

Use standalone or combine them. Start simple, add intelligence as you need it.

### Core Products

**[Vector Memory](./products.md#vector-memory)** - Semantic search and RAG foundation  
**[Chat Memory](./products.md#chat-memory)** - Conversation storage with compression  
**[Document Intelligence](./products.md#document-intelligence)** - Extract from PDFs and docs  
**[AI Model Proxy](./products.md#ai-model-proxy)** - Unified multi-model API

### Add Intelligence

**[Knowledge Graphs](./products.md#knowledge-graphs)** - Map entity relationships (addon)  
**[Graph-Aware Search](./products.md#graph-aware-search)** - Domain-tuned retrieval (addon)  
**[Sync & Portability](./products.md#sync--portability)** - Local/cloud sync (feature)

[Compare all products →](./products.md)

---

## Data In, Intelligence Out

**Three ways to send data:**
1. Documents (PDFs, Word docs) - `POST /v1/document`
2. Conversations (chat messages) - `POST /v1/messages`
3. Direct memories (explicit data) - `POST /v1/memory`

**Intelligence layer transforms it:**
- Vector embeddings (semantic understanding)
- Knowledge graphs (relationship mapping)
- Predictive caching (anticipate needs)
- Domain tuning (specialized search)

**Two ways to query:**
1. Natural language search - `POST /v1/memory/search`
2. GraphQL analytics - `POST /v1/graphql`

[See architecture →](../concepts/architecture.md) | [API reference →](../apis/index.yaml)

---

## What You Can Build

[Personal AI Assistant](../tutorials/chat-history.md) - Store/retrieve conversations  
[Document Q&A](../tutorials/document-qa.md) - Intelligent document chat  
[Customer Support](../tutorials/customer-experience.md) - Answer FAQs, resolve tickets  
[Enterprise Knowledge](../tutorials/enterprise-saas.md) - Multi-tenant intelligence  
[Code Search](../guides/graph-aware-embeddings.md) - Find code by natural language  
[Domain Ontologies](../guides/custom-schemas.md) - Custom knowledge graphs  
[Graph Analytics](../guides/graphql-analysis.md) - Query insights with GraphQL

---

## Start Here

### 1. Choose Your Path

**Building something specific?**
- [Document Q&A](../tutorials/document-qa.md) - Extract + search PDFs
- [Conversational AI](../tutorials/chat-history.md) - Chat with memory
- [Code Search](../guides/graph-aware-embeddings.md) - Find code by intent
- [Knowledge Management](../tutorials/enterprise-saas.md) - Multi-tenant intelligence

**Just exploring?**
- [Products Overview](./products.md) - See all seven products
- [Quick Start (5 min)](../quickstart/index.md) - Ship a prototype
- [Decision Tree](./decision-tree.md) - Which products do you need?

### 2. Integrate

- [TypeScript SDK](../sdks/typescript.md) | [Python SDK](../sdks/python.md)
- [REST API Reference](../apis/index.yaml)
- [LangChain / LlamaIndex Examples](../examples/index.md)

---

## Deployment Options

**Papr Cloud** - Managed service, 5-minute setup  
[Get started →](../quickstart/index.md) | [Learn more →](../deployment/cloud.md)

**Hybrid Cloud** - Your infrastructure, we manage it  
[Enterprise →](../deployment/hybrid.md) | [Talk to sales →](https://calendly.com/amirkabbara/30min)

**Self-Hosted** - Open source, full control  
[Setup →](../deployment/self-hosted.md) | [GitHub →](https://github.com/Papr-ai/memory-opensource)

All deployment options use **identical APIs**. Code written for one works with all three.

[Compare deployments →](../deployment/index.md)

---

## Why Teams Choose Papr

**Instead of building:**
- Basic RAG (70-80% accuracy)
- Manual fusion of keyword + vector
- Custom knowledge graphs
- Fragmented data sources

**You get:**
- 91%+ retrieval accuracy (#1 benchmark)
- Hybrid search built-in
- Predictive intelligence layer
- Connected context across sources
- Seven products, one API
- Open source + enterprise

[See detailed comparison →](./why-papr.md)