Fleetlit Roadmap

Supabase for AI Era - Product Roadmap

Vision

Build a comprehensive Backend-as-a-Service platform specifically designed for AI applications, providing the essential infrastructure that AI/SaaS applications need.

Current Features ✅

Core Infrastructure

✅ Multi-tenant Architecture - Users, Teams, Projects
✅ User Management - Authentication, JWT, Refresh Tokens
✅ File Storage - S3-compatible, Cloudflare R2
✅ Event-Driven - Kafka-based
✅ Asset Exchange - AssetFlow framework
✅ Runtime Variables - Dynamic configuration
✅ Collections & Threads - Data organization

Roadmap by Priority

🚀 Phase 1: Essential AI/SaaS Features (High Priority)

1. Realtime / 实时通信 🔴

Status: Planned

Effort: 3-4 weeks

AI applications need real-time capabilities for:

Live chat and AI assistant interactions
Real-time collaboration features
Live updates and notifications

Implementation:

WebSocket server integration
Real-time database subscription (PostgreSQL Logical Replication)
Presence and online status tracking
Room-based messaging for multi-user scenarios
Event broadcasting system

Tech Stack:

WebSocket (Go: gorilla/websocket)
Redis Pub/Sub for horizontal scaling
PostgreSQL NOTIFY/LISTEN
Connection management and reconnection handling

API Endpoints:

WS /realtime/v1/subscribe        - WebSocket connection
GET /realtime/v1/presence         - Get online users
POST /realtime/v1/broadcast       - Broadcast message

2. Notification System / 通知系统 🔴

Status: Planned

Effort: 3-4 weeks

Multi-channel notifications for user engagement:

Email notifications (SMTP/SendGrid)
SMS notifications (Twilio)
Push notifications (Firebase/APNs)
In-app notifications
Webhook notifications

Implementation:

Notification service with queue-based delivery
Template management for different notification types
User notification preferences
Notification history and delivery status
Rate limiting and cost control

Tech Stack:

Kafka for async delivery
SMTP provider (SendGrid/Mailgun)
SMS provider (Twilio)
Push service (Firebase Cloud Messaging)

3. Task Queue & Background Jobs / 任务队列 🔴

Status: Planned

Effort: 3-4 weeks

Essential for async processing in AI applications:

AI model inference tasks
Batch data processing
Scheduled jobs (cron-like)
Long-running background tasks
Retry mechanism with exponential backoff
Dead letter queue

Implementation:

Task queue system (Redis-based or RabbitMQ)
Worker pool management
Task scheduling and cron jobs
Task status tracking
Priority queues
Task chaining and workflows

API Endpoints:

POST /tasks/v1/create           - Create task
GET  /tasks/v1/:id              - Get task status
POST /tasks/v1/:id/cancel       - Cancel task
GET  /tasks/v1/:id/result       - Get task result

4. Subscription & Billing / 订阅与计费 🔴

Status: Planned

Effort: 4-5 weeks

Essential for SaaS business model:

Subscription plan management
Payment processing (Stripe, PayPal, Alipay, WeChat Pay)
Usage-based billing and quota management
Invoice generation and management
Trial and promotion management

Payment Providers:

Stripe (Global cards, international)
PayPal (Global)
Alipay (China)
WeChat Pay (China)

Key Features:

Flexible Billing Models
- Subscription-based: Recurring charges
- Usage-based: Pay-as-you-go
- Hybrid: Base + usage overages
- Tiered pricing
Subscription Lifecycle
- Trial periods
- Automatic renewal
- Grace periods
- Cancellation (immediate or end of period)
- Plan changes (upgrade/downgrade)
AI-Specific Features
- Token quotas per plan
- API rate limits per tier
- Storage quotas
- Feature-based pricing

API Endpoints:

# Plans
GET  /billing/v1/plans              - List plans
POST /billing/v1/plans              - Create plan (admin)
GET  /billing/v1/plans/:id          - Get plan details

# Subscriptions
POST /billing/v1/subscriptions      - Create subscription
GET  /billing/v1/subscriptions      - List subscriptions
PUT  /billing/v1/subscriptions/:id  - Update (upgrade/downgrade)
POST /billing/v1/subscriptions/:id/cancel - Cancel subscription

# Payment Methods
POST /billing/v1/payment_methods     - Add payment method
GET  /billing/v1/payment_methods     - List payment methods
DELETE /billing/v1/payment_methods/:id - Remove payment method

# Invoices
GET  /billing/v1/invoices           - List invoices
GET  /billing/v1/invoices/:id       - Get invoice
GET  /billing/v1/invoices/:id/pdf   - Download invoice PDF

🎯 Phase 2: AI-Specific Features (Medium Priority)

5. Vector Database & Embeddings / 向量数据库 🟡

Status: Planned

Effort: 4-5 weeks

Core infrastructure for AI applications:

Vector storage and indexing
Similarity search (cosine, Euclidean)
Vector operations and transformations
Integration with embedding providers
Hybrid search (vector + keyword)

Implementation:

Vector extension for PostgreSQL (pgvector)
Vector similarity search API
Embedding generation service
Vector indexing strategies
Batch embedding operations

API Endpoints:

POST /vectors/v1/upsert         - Insert/update vectors
POST /vectors/v1/search         - Similarity search
GET  /vectors/v1/:id            - Get vector
DELETE /vectors/v1/:id          - Delete vector
POST /vectors/v1/embed          - Generate embeddings

6. AI Model Integration / AI 模型集成 🟡

Status: Planned

Effort: 3-4 weeks

Unified API for multiple AI providers:

LLM providers (OpenAI, Anthropic, Cohere)
Embedding providers
Image generation providers
Model routing and load balancing
Cost tracking and quota management

Providers:

OpenAI (GPT-4, GPT-3.5, Embeddings)
Anthropic (Claude)
Cohere (Command, Embed)
Azure OpenAI
Local models (Ollama)

API Endpoints:

POST /ai/v1/completions        - Text completion
POST /ai/v1/chat              - Chat completion
POST /ai/v1/embeddings        - Generate embeddings
POST /ai/v1/images/generate   - Image generation
GET  /ai/v1/models            - List available models

7. Prompt Management / Prompt 管理 🟡

Status: Planned

Effort: 2-3 weeks

Manage and version prompts for AI applications:

Prompt templates with variables
Version control and A/B testing
Prompt performance analytics
Collaboration and sharing
Prompt testing playground

API Endpoints:

POST /prompts/v1/create        - Create prompt
GET  /prompts/v1/:id           - Get prompt
PUT  /prompts/v1/:id           - Update prompt
POST /prompts/v1/:id/test      - Test prompt
GET  /prompts/v1/:id/versions  - List versions

8. Conversation & Memory Management / 对话与记忆 🟡

Status: Planned

Effort: 3-4 weeks

Support for AI chat applications:

Conversation history storage
Message threading
Context window management
Memory summarization
RAG (Retrieval Augmented Generation) support

API Endpoints:

POST /conversations/v1/create   - Create conversation
GET  /conversations/v1/:id      - Get conversation
POST /conversations/v1/:id/send - Send message
GET  /conversations/v1/:id/memory - Get context memory

🔧 Phase 3: Developer Experience & DevOps (Medium Priority)

9. Auto-generated REST API / 自动生成 REST API 🔵

Status: Planned

Effort: 2-3 weeks

Generate RESTful APIs from database schema:

CRUD operations for all tables
Filtering, sorting, pagination
Relationship handling
Row Level Security (RLS)
API authentication

Inspired by: PostgREST, Supabase API

API Pattern:

GET    /tables/:table           - List records
POST   /tables/:table           - Create record
GET    /tables/:table/:id       - Get record
PUT    /tables/:table/:id       - Update record
DELETE /tables/:table/:id       - Delete record

10. Row Level Security (RLS) / 行级安全 🔵

Status: Planned

Effort: 3-4 weeks

Fine-grained access control at database level:

Policy-based access control
User/team/project isolation
Dynamic policy evaluation
Policy inheritance
Performance optimization

Example Policies:

-- Users can only see their own data
CREATE POLICY user_isolation ON users
  FOR ALL USING (user_id = current_user_id());

-- Team members can access team data
CREATE POLICY team_access ON projects
  FOR ALL USING (
    team_id IN (
      SELECT team_id FROM team_members
      WHERE user_id = current_user_id()
    )
  );

11. API Key Management / API 密钥管理 🔵

Status: Planned

Effort: 2 weeks

API key system for external integrations:

API key generation and management
Key types (public, secret, publishable)
Key permissions and scopes
Usage quotas and rate limits
Key expiration and rotation

API Endpoints:

POST /keys/v1/create           - Create API key
GET  /keys/v1                  - List keys
DELETE /keys/v1/:id            - Delete key
POST /keys/v1/:id/rotate       - Rotate key

12. Webhooks / Webhook 支持 🔵

Status: Planned

Effort: 2-3 weeks

Webhook system for event notifications:

Webhook registration and management
Event type subscription
Retry mechanism with backoff
Signature verification (HMAC)
Delivery status tracking

API Endpoints:

POST /webhooks/v1/create       - Create webhook
GET  /webhooks/v1              - List webhooks
DELETE /webhooks/v1/:id        - Delete webhook
GET  /webhooks/v1/:id/logs     - Delivery logs

📊 Phase 4: Monitoring & Analytics (Lower Priority)

13. Analytics & Insights / 分析与洞察 🟢

Status: Planned

Effort: 4-5 weeks

Built-in analytics for applications:

Event tracking and analytics
User behavior analytics
Funnel analysis
Retention analysis
Custom dashboards
Export and reporting

API Endpoints:

POST /analytics/v1/track       - Track event
GET  /analytics/v1/query       - Query analytics
GET  /analytics/v1/funnel      - Funnel analysis
GET  /analytics/v1/retention   - Retention analysis

14. Logging & Monitoring / 日志与监控 🟢

Status: Planned

Effort: 3-4 weeks

Comprehensive observability:

Structured logging
Log aggregation and search
Metrics collection (Prometheus)
Distributed tracing (OpenTelemetry)
Alert management
Performance monitoring

Tech Stack:

Loki (log aggregation)
Prometheus (metrics)
Grafana (dashboards)
OpenTelemetry (tracing)

🛡️ Phase 5: Security & Compliance (Lower Priority)

16. Rate Limiting / 速率限制 🟢

Status: Planned

Effort: 2 weeks

Protect API from abuse:

Request rate limiting
Concurrent request limiting
Per-user/per-key limits
Sliding window algorithm
Distributed rate limiting

17. Audit Logging / 审计日志 🟢

Status: Planned

Effort: 2-3 weeks

Comprehensive audit trail:

Data access logging
Data modification tracking
User activity logging
Compliance reporting
Log export and archival

18. Data Encryption / 数据加密 🟢

Status: Planned

Effort: 2-3 weeks

Enhanced data security:

Field-level encryption
Encryption at rest
Encryption in transit
Key management
PCI DSS compliance support

🌐 Phase 6: Edge & Performance (Future)

19. Edge Functions / 边缘函数 🟢

Status: Planned

Effort: 4-5 weeks

Serverless functions at the edge:

Function deployment
Multi-region deployment
Edge caching
Function scheduling
Version management

Tech Stack:

WebAssembly (Wasm)
Cloudflare Workers integration
Edge runtime

20. CDN Integration / CDN 集成 🟢

Status: Planned

Effort: 2 weeks

Global content delivery:

Static asset CDN
API response caching
Image optimization
Video streaming
Multi-region deployment

Implementation Timeline

Q1 2025

✅ Asset Flow optimization
✅ Logout implementation
🚀 Realtime / 实时通信
🚀 Notification System / 通知系统
🚀 Task Queue / 任务队列
🚀 Subscription & Billing / 订阅与计费

Q2 2025

🎯 Vector Database / 向量数据库
🎯 AI Model Integration / AI 模型集成
🎯 Prompt Management / Prompt 管理
🎯 Conversation & Memory / 对话与记忆

Q3 2025

🔧 Auto-generated REST API / 自动生成 REST API
🔧 Row Level Security / 行级安全
🔧 API Key Management / API 密钥管理
🔧 Webhooks / Webhook 支持

Q4 2025

📊 Analytics / 分析
📊 Logging & Monitoring / 监控与日志
🛡️ Rate Limiting / 速率限制
🛡️ Audit Logging / 审计日志

Contributing

This roadmap is a living document. We welcome:

Feature requests
Priority feedback
Implementation contributions
Bug reports and fixes

Please open an issue or PR to discuss roadmap changes.

Last Updated: 2025-01-08