Skip to main content

Roadmap Overview

This section documents planned features and enhancements for Cortex. These pages help you understand future capabilities and plan your integrations accordingly.

Disclaimer

All features listed in this roadmap are planned but not yet implemented. Availability, specifications, and timelines are subject to change. We recommend using current features (see Core Features) for production applications.

Planned Features by Category​

Infrastructure & Services​

Cloud Mode

Managed services with auto-embeddings, GDPR compliance, and infrastructure

MCP Server

Cross-application memory via Model Context Protocol

Cloud Mode provides enterprise-ready managed services including automatic embeddings, GDPR-compliant cascade deletion, managed pub/sub infrastructure, analytics dashboards, and automatic governance enforcement. Your data stays in your Convex instance while Cortex Cloud handles operational complexity.

MCP Server enables memory sharing across AI tools like Cursor, Claude Desktop, and custom applications. Tell one tool something, all tools remember it.

Core Capabilities​

Context Chains

Advanced workflow management capabilities

A2A Communication

Real-time subscriptions and smart routing

Infinite Context

Advanced retrieval strategies

Context Chains enhancements include status transition validation, auto-complete parent contexts, automatic data inheritance, chain-wide participant propagation, and depth limit enforcement.

A2A Communication enhancements include real-time subscriptions via WebSocket, convenience inbox methods (getInbox(), getSent(), markRead()), Cloud Mode managed pub/sub, automated agent execution via webhooks, analytics dashboards, smart routing, and automated summarization.

Infinite Context enhancements include predictive pre-fetching, hierarchical retrieval (coarse-grained then fine-grained), cross-MemorySpace retrieval, adaptive fact extraction, automatic hybrid search, and content summarization.

Platform & APIs​

API Enhancements

Bulk operations, filtering, and exports

Governance

Automatic policy enforcement

Access Analytics

Memory usage and performance analytics

API Enhancements include dry-run modes for bulk operations, confirmation thresholds, advanced filtering (text search, nested queries, range operators), enhanced export capabilities (direct file writing, version history, linked conversations), multi-tenancy filters, and sessions API improvements.

Governance enhancements include automatic policy enforcement on every storage operation, automatic version purging, sessions policy enforcement, scheduled enforcement jobs, enforcement hooks, policy validation, and enforcement metrics.

Access Analytics provides the full cortex.analytics.* API with visual dashboards for memory growth, search performance, access heat maps, predictive analytics, cost tracking, and automated alerts.

Feature Categories​

High Priority​

Features that significantly improve developer experience or eliminate operational complexity:

  • Automatic Policy Enforcement (Governance) - No more manual enforce() calls
  • Convenience Inbox Methods (A2A) - getInbox(), getSent(), markRead()
  • Dry Run Mode (API) - Preview bulk operations before executing
  • Status Transition Validation (Context Chains) - Prevent invalid state changes
  • Automatic Hybrid Search (Infinite Context) - Simplified semantic + keyword search

Medium Priority​

Features that add flexibility and power for advanced use cases:

  • Hierarchical Retrieval (Infinite Context) - Scalability for large memory spaces
  • Smart Routing (A2A) - AI-powered agent selection
  • Advanced Filtering (API) - Nested queries, range operators
  • Automatic Data Inheritance (Context Chains) - Simplified context chain data management

Cloud Mode (Enterprise)​

Enterprise features requiring managed infrastructure:

  • Auto-Embeddings - Zero-config embedding generation
  • GDPR Cascade Deletion - One-click compliance with legal guarantees
  • Managed Pub/Sub - Real-time A2A without managing Redis/NATS
  • Analytics Dashboards - Visual insights for memory, agents, and communication
  • Agent Billing & Limits - Usage tracking and resource limits

Migration Path​

All planned features are designed to be backward compatible. Existing code will continue to work unchanged when new features are released.

Example: Enabling Cloud Mode​

// Before (Direct Mode)
const cortex = new Cortex({
  convexUrl: process.env.CONVEX_URL!,
});

// After (Cloud Mode) - just add API key
const cortex = new Cortex({
  convexUrl: process.env.CONVEX_URL!,
  apiKey: "cortex_sk_...",
  mode: "cloud",
  autoEmbed: true, // Optional: automatic embeddings
});

Example: Opting Into New Features​

// Existing code (works unchanged)
await cortex.contexts.update(spaceId, contextId, {
  status: "completed",
});

// New features (opt-in)
await cortex.contexts.update(spaceId, contextId, {
  status: "completed",
  validateTransition: true, // Opt-in to status validation
});

Current vs. Planned​

FeatureCurrent StatusPlanned Enhancement
EmbeddingsManual generation requiredAuto-embeddings (Cloud Mode)
GDPR DeletionManual cascade deletionOne-click with legal guarantees (Cloud Mode)
A2A MessagingManual polling or manual pub/subReal-time subscriptions (Cloud Mode)
GovernanceManual enforce() callsAutomatic enforcement
Bulk OperationsDirect executionDry-run mode, confirmation thresholds
FilteringBasic filtersNested queries, range operators, text search
Context ChainsManual status managementAutomatic validation, inheritance, completion
SearchSemantic or keyword (separate)Automatic hybrid search
AnalyticsBasic stats via getStats()Full analytics dashboard (Cloud Mode)

Workarounds​

Until planned features are available, current workarounds are documented in each roadmap page:

Stay Updated​

For the latest updates on feature development:

Feedback​

Have ideas for features not listed here? We'd love to hear from you:

  • GitHub Discussions: Share feature ideas and vote on proposals
  • GitHub Issues: Report bugs or request specific enhancements