TMRW Venture Factory

TOMORROW INC

Launch 1000 Ships — Idea → Product → Company in hours

📂 Structure

      

        
🛠 Stack

        
🧠 Context Files

      

    

  


  
  

    

      
1. Ideate

        
1.Edward defines venture concept

        
2.Agents research market + competitors

        
3.Generate SYSTEM_PROMPT.md

        
4.Define API contracts + data schemas

      

      
2. Scaffold

        
1.tmrw create <type> <name>

        
2.Terraform provisions ECS + RDS + ALB

        
3.GitHub repo + CI/CD auto-configured

        
4.Stripe + SES + RDS wired up

      

      
3. Build

        
1.AI agents build using context docs

        
2.Animesh reviews infra + deploys

        
3.Ronnie drives GTM + design system

        
4.Zach wires operational tooling

      

      
4. Launch

        
1.make deploy-prod

        
2.ECS service auto-scales

        
3.Lead capture + email flows live

        
4.Monitor via CloudWatch + /health

      

    

    
⌨ TMRW CLI Commands

      

        
tmrw create microapp myappScaffold a microapp

        
tmrw create marketplace mympScaffold a marketplace

        
tmrw create saas mysaasScaffold a SaaS product

        
tmrw create service mysvcScaffold a service company

        
tmrw create api myapiScaffold an API service

        
tmrw create mobile myappScaffold a mobile app

        
tmrw create pipeline mypipeScaffold a data/ML pipeline

        
tmrw create fullstack myprojScaffold full-stack web app

        
make devStart local dev environment

        
make deploy-stagingDeploy to staging ECS

        
make deploy-prodDeploy to production ECS

        
make infra-upProvision AWS via Terraform

        
make ecs-statusCheck ECS service health

        
make logsTail CloudWatch logs

        
make scale <count>Scale ECS desired count

        
make rollbackRollback to previous task def

      

    

  


  
  

    

      
🧠 Context Engineering for "Launch 1000 Ships"

      
Every venture TMRW launches gets context documents that serve as the knowledge layer for both the team and AI agents. When an agent builds a feature, it loads the venture's SYSTEM_PROMPT.md + relevant architecture docs. When Animesh deploys infrastructure, the RUNBOOK.md provides exact procedures. This is how you make agentic operations reliable at scale.

    

    

      
# Venture: [Name]
# Type: microapp | marketplace | saas | service | api | fullstack | mobile | pipeline
# Owner: Edward Unthank / Tomorrow Inc

## Domain
What this venture does, who it serves, core value proposition.

## TMRW Stack Conventions
- Framework: Next.js 14+ (App Router) | FastAPI | Go | React Native
- Styling: TailwindCSS
- Validation: Zod schemas for all inputs
- ORM: Prisma → PostgreSQL (AWS RDS)
- Email: AWS SES (PENDING → SENT → FAILED)
- Payments: Stripe (subscriptions, checkout, portal)

## Infrastructure (ECS)
- Compute: ECS Fargate (auto-scaling)
- Load Balancer: ALB with health checks
- Registry: ECR for Docker images
- Database: RDS PostgreSQL (per-venture schema)
- DNS: Route53 → ALB → ECS
- SSL: ACM certificates (auto-renewed)

## Code Conventions
- TypeScript strict mode
- Zod schema validates before DB write
- API responses: { success, data, error }
- Health endpoint required: /health and /api/health
- Docker multi-stage builds for small images

## What NOT to Do
- Never hardcode secrets (use AWS Secrets Manager + .env)
- Never skip Zod validation on API routes
- Never deploy without staging test first
- Never modify shared Terraform without Animesh review
- Never set ECS desired count to 0 in prod

    

    

      
# Architecture: [Venture Name]

## Shared Infrastructure (managed by Animesh)
- AWS Account: coaip | Region: us-east-1
- ECS Cluster: tmrw-ventures (Fargate)
- ALB: tmrw-alb (shared, path/host-based routing)
- ECR: per-venture repositories
- RDS PostgreSQL (shared cluster, per-venture schema)
- SES email (verified domain)
- Route53 DNS (*.tmrwgroup.ai)
- Secrets Manager (per-venture secrets)

## ECS Service Architecture
┌──────────────────────────────────────────┐
│  Route53: venture.tmrwgroup.ai           │
│              ↓                           │
│  ALB (shared) → Target Group             │
│              ↓                           │
│  ECS Service (Fargate)                   │
│  ├── Task Definition                     │
│  │   ├── App Container (Next.js/API)     │
│  │   └── Sidecar (log router, optional)  │
│  ├── Desired Count: 2 (prod) / 1 (stg)  │
│  ├── Auto-scaling: CPU/Memory targets    │
│  └── Health Check: /health               │
│              ↓                           │
│  RDS PostgreSQL (venture_db schema)      │
│  Secrets Manager (API keys, DB creds)    │
│  SES (transactional email)               │
│  Stripe (payments)                       │
└──────────────────────────────────────────┘

## Deployment
GitHub push → Actions CI → Docker build → ECR push
→ ECS task definition update → rolling deployment
→ ALB health check passes → traffic shifts → live

## Scaling
- Min: 1 task (staging), 2 tasks (production)
- Max: 10 tasks
- Target: 70% CPU, 80% memory
- Scale-in cooldown: 300s

    

    

      
# Agent Task → Context Loaded

"Build the landing page for VentureX"
→ SYSTEM_PROMPT.md + DESIGN_SYSTEM.md + /pages examples

"Set up Stripe billing for VentureX"
→ SYSTEM_PROMPT.md + API_CONTRACTS.md + Stripe docs

"Deploy VentureX to production"
→ SYSTEM_PROMPT.md + ARCHITECTURE.md + RUNBOOK.md

"Scale VentureX ECS service"
→ ARCHITECTURE.md + RUNBOOK.md + Terraform modules

"Add a new ML model to the pipeline"
→ SYSTEM_PROMPT.md + DATA_DICTIONARY.md + EXPERIMENT_LOG.md

"Build a React Native screen"
→ SYSTEM_PROMPT.md + DESIGN_SYSTEM.md + navigation config

"Create a new API endpoint"
→ SYSTEM_PROMPT.md + API_CONTRACTS.md + existing routes

"Debug a failing ECS deployment"
→ ARCHITECTURE.md + RUNBOOK.md + CloudWatch logs

# The Pattern:
# 1. Always load SYSTEM_PROMPT.md (venture-specific)
# 2. Add task-relevant architecture/contract docs
# 3. Add relevant source files as examples
# 4. Agent works with full context → reliable output

    

    

      
# Living Website: Context-Driven Auto-Updates

## Trigger Events
- New team member → agents update /team page
- New venture launched → agents update homepage + /ventures
- New content published → agents update blog/news
- Founding customer signed → agents update social proof

## How It Works
1. Event fires (webhook, DB trigger, Slack message)
2. Agent loads: SYSTEM_PROMPT.md + DESIGN_SYSTEM.md + current page
3. Agent generates updated component/content
4. PR created automatically → CI runs
5. Staging preview → auto-merge if tests pass
6. ECS rolling deploy → production updates

## ECS Advantage
- Zero-downtime deploys via rolling update
- Health check gates traffic to new version
- Instant rollback if deploy fails

    

  


  
  

    

      
☁ TMRW Infrastructure (ECS)

        

          
Cloud ProviderAWS (coaip)

          
Regionus-east-1

          
ComputeECS Fargate

          
Load BalancerALB (shared)

          
Container RegistryECR

          
DatabaseRDS PostgreSQL

          
DNSRoute53

          
EmailAWS SES

          
SecretsSecrets Manager

          
SSLACM (auto-renew)

          
MonitoringCloudWatch

          
Productiontmrwgroup.ai

          
Stagingstaging.tmrwgroup.ai

        

      

      
🏗 ECS Terraform Module

        
# tmrw create microapp "ventureX"
# provisions all of this via Terraform:

module "ventureX" {
  source = "./modules/ecs-venture"

  name        = "ventureX"
  domain      = "venturex.tmrwgroup.ai"
  environment = "production"

  # ECS
  cluster_arn   = aws_ecs_cluster.tmrw.arn
  cpu           = 256       # 0.25 vCPU
  memory        = 512       # 512 MB
  desired_count = 2
  max_count     = 10

  # Auto-scaling
  cpu_target    = 70
  memory_target = 80

  # ALB
  alb_arn          = aws_lb.tmrw.arn
  listener_arn     = aws_lb_listener.https.arn
  health_check_path = "/health"

  # Database
  db_schema     = "venturex_db"
  rds_cluster   = aws_rds_cluster.tmrw.id

  # Services
  ses_domain    = "tmrwgroup.ai"
  route53_zone  = aws_route53_zone.tmrw.id
  acm_cert_arn  = aws_acm_certificate.wild.arn

  # Payments
  stripe_enabled = true
}

# Output:
# → ECS service running (Fargate)
# → ALB target group + routing rule
# → ECR repository created
# → Database schema provisioned
# → DNS records pointing (A + AAAA)
# → SSL via ACM
# → CI/CD pipeline configured
# → Auto-scaling policies active
# → CloudWatch alarms set
# → Secrets Manager entries created

      

    


    
    
🏛 ECS Architecture

      

        

          
Edge / DNS

          

            Route53 (*.tmrwgroup.ai)
            ACM SSL Certificates
          

        

        
↓

        

          
Load Balancing

          

            ALB (shared)
            Host-based routing
            Target groups per venture
            Health checks
          

        

        
↓

        

          
Compute — ECS Fargate Cluster

          

            Service: tmrwgroup-site (2 tasks)
            Service: ventureX (2 tasks)
            Service: ventureY (1 task)
            Auto-scaling policies
            Rolling deployments
            ECR image registry
          

        

        
↓

        

          
Data + Services

          

            RDS PostgreSQL (shared cluster)
            Per-venture schemas
            AWS SES (email)
            Stripe (payments)
            Secrets Manager
            CloudWatch (logs + metrics)
          

        

      

    


    
    
📊 Deployment Pipeline (ECS)

      

        
📝
Git Push

        →
        
✅
CI Tests

        →
        
🐳
Docker Build

        →
        
📦
ECR Push

        →
        
🔄
ECS Deploy

        →
        
🌐
Live

      

    

  


  

    TMRW Group · Tomorrow Inc · "Launch 1000 Ships"
    Ideate → Scaffold → Build → Launch