Homepage Reimagined Spec
This spec defines a developer-first homepage narrative and production-ready copy for index.page.tsx.
Goal: get developers excited and oriented in one pass:
- understand what Papr is,
- trust the claims,
- run first success fast,
- see how Papr scales with their stack.
Core Principles
- Lead with
whatbeforewhy. - Keep claims specific, source-backed, and consistent with docs.
- Show first success immediately with runnable examples.
- Map capabilities to endpoints and integration paths.
- Avoid competitor callouts; focus on developer outcomes.
Page Flow (Top to Bottom)
- Hero (What Papr is)
- First Success in 60 Seconds (store + search)
- Choose Your Path (4 implementation tracks)
- Why This Architecture (keyword -> vector -> graph -> predictive)
- Capabilities -> API Surface (clear endpoint mapping)
- Integrations and Frameworks (meet developers where they are)
- Trust and Deployment (security + deployment parity)
- Final CTA (start now)
Exact Copy
1) Hero
Pill text
#1 on STaRK retrieval benchmark
Headline
Memory layer for AI applications
Subheadline
Unify chat, documents, and structured data into a predictive memory graph. Start simple with store and search, then scale to graph-aware retrieval and predictive context without rewrites.
Proof row
91%+ retrieval accuracy on STaRK<150ms retrieval when cached96% conversation compression for long sessions
Primary CTA
Run Quick Start
Secondary CTA
Explore Integration Paths
CTA targets:
- Primary ->
/quickstart - Secondary ->
/overview/golden-paths
2) First Success in 60 Seconds
Section title
First success in 60 seconds
Section subtitle
Store one memory, retrieve it with graph-aware search, and verify context reuse.
Code tabs
cURLPythonTypeScript
Validation text under snippet
Expected result: a successful write response, then search results containing user communication preferences.
3) Choose Your Path
Section title
Choose your integration path
Section subtitle
Pick the starting point that matches your architecture.
Card 1
- Title:
Add memory to existing agent - Description:
Retrofit retrieval-before-response and writeback-after-response into your current agent loop. - CTA:
Use this path - Link:
/quickstart/add-memory-to-existing-agent
Card 2
- Title:
Chat memory - Description:
Use sessioned Messages API with compression for multi-turn conversations. - CTA:
Use this path - Link:
/quickstart/chat-memory
Card 3
- Title:
Document memory - Description:
Ingest PDFs and Word docs and retrieve grounded context across sessions. - CTA:
Use this path - Link:
/quickstart/document-memory
Card 4
- Title:
Structured data memory - Description:
Map operational data (for example Postgres rows) into graph nodes and relationships. - CTA:
Use this path - Link:
/quickstart/structured-data-memory
4) Why This Architecture
Section title
Why this architecture works in production
Section subtitle
From simple retrieval to predictive memory without replacing your stack.
Progression blocks
Keyword search- exact matches, good for narrow vocab.Vector search- semantic recall across phrasing variation.Graph retrieval- relationship-aware multi-hop context.Predictive memory- pre-ranked, behavior-aware retrieval that improves with usage.
Supporting sentence
Papr combines all four layers behind one API, so teams can start simple and gain sophistication as data grows.
5) Capabilities Mapped to APIs
Section title
Capabilities mapped to APIs
Capability list:
Store explicit memory->POST /v1/memoryRetrieve with agentic graph->POST /v1/memory/searchCompress long chat sessions->POST /v1/messages/context_for_llmIngest documents->POST /v1/documentGraph analytics->POST /v1/graphqlSync and portability->POST /v1/sync/tiers,POST /v1/sync/delta,POST /v1/omo/export,POST /v1/omo/import
CTA:
View API Reference -> /apis
6) Integrations and Frameworks
Section title
Integrates with the tools developers already use
Section subtitle
Use Papr as the memory layer in your current stack.
Integration tiles (link targets):
MCP->/integrations/mcp-serverOpenClaw->/integrations/openclawLangGraph->/integrations/langgraphCrewAI->/integrations/crewaiAutoGen->/integrations/autogenVercel AI SDK->/examples/vercel-ai-sdk-papr
7) Trust and Deployment
Section title
Security, isolation, and deployment flexibility
Bullets:
ACL-aware memory policy controls read and write boundaries.Organization and namespace isolation for multi-tenant systems.Cloud, hybrid, and self-hosted deployment with the same API surface.
CTA:
Compare deployment modes -> /deployment
8) Final CTA
Title
Build your first memory-powered workflow today
Buttons:
Get API Key->https://dashboard.papr.aiOpen Quick Start->/quickstart
Component Mapping for index.page.tsx
Use this mapping to update current sections without a full rewrite.
- Current hero (
Hero,RankingPill,Typography) -> replace copy with Hero section above. - Current
CardWithCodeblock -> convert to "First success in 60 seconds"; keep tabs but use spec-compliant payloads. - Current
use-casesandtemplatessections -> replace with "Choose your integration path" cards. - Current
WorkflowSection-> retitle and reframe as "Why this architecture works in production". - Current dashboard section -> repurpose into "Capabilities mapped to APIs".
- Keep bottom CTA container, update copy and button targets.
Code Example Guardrails (Must Follow)
- Put
external_user_idat request top level. - Do not place
user_idorexternal_user_idinsidemetadata. - Use
metadata.customMetadatafor arbitrary custom fields. - Keep
customMetadatavalues flat: string, number, boolean, or string array. - Prefer
memory_policy.aclover legacy metadata access fields.
Success Metrics
Track these post-launch:
- Hero-to-QuickStart CTR.
- API key clickthrough rate from homepage.
- First successful API call completion rate.
- Integration guide entry rate from homepage.
- Time-to-first-success (first write + first search).
Implementation Order
- Replace hero copy and CTA labels.
- Update code snippets to spec-compliant payloads.
- Replace templates/use-cases with "Choose Your Path".
- Reframe workflow section to architecture progression.
- Add capabilities-to-endpoint mapping block.
- Add integrations strip and deployment trust section.
- QA all links and run docs example validation.