Skip to content

Tax Practice AI - Next Steps

Last updated: 2024-12-23

Current Status: Development Environment Setup (Complete)

Completed

  • Virtual environment created (.venv/)
  • Python dependencies installed (requirements.txt, requirements-dev.txt)
  • docker-compose.yml for local PostgreSQL (port 5433)
  • src/ directory structure created
  • tests/ directory structure created
  • .env.example template
  • .gitignore updated
  • config.yaml with documented parameters
  • .env configured with DB_* variables
  • PostgreSQL container running (tax-practice-postgres)
  • DATABASE_SCHEMA.sql applied (48 tables)
  • Base service classes created:
  • src/config/settings.py (config loading with env var substitution)
  • src/services/base_service.py (abstract base class)
  • src/services/aurora_service.py (async PostgreSQL access)
  • src/services/init.py (ServiceRegistry)

Verified Working

# Test database connectivity
source .venv/bin/activate && source .env
python3 -c "
import asyncio
from src.config import get_config
from src.services import services

async def test():
    config = get_config()
    services.initialize(config)
    result = await services.aurora.health_check_async()
    print(f'Database connected: {result}')
    await services.close_all()

asyncio.run(test())
"

Architectural Decision: Multi-Tenant SaaS

Decision: Separate Databases per Tenant (within one Aurora cluster)

Tenant Database Connection
Firm A tax_practice_firma aurora-cluster/tax_practice_firma
Firm B tax_practice_firmb aurora-cluster/tax_practice_firmb

Rationale: - Strongest data isolation (critical for tax/compliance) - Single Aurora cluster = shared tiered pricing - Independent backups per tenant - Easy migration to dedicated cluster if tenant grows - Meets IRS Circular 230 and WISP requirements

Implementation requirements: - [ ] Tenant registry table (master database or separate service) - [ ] Tenant routing middleware (resolve tenant → connection) - [ ] Per-tenant database provisioning workflow - [ ] Tenant-aware ServiceRegistry (dynamic connections)

Implementation Workflow

Before Coding Any Feature

  1. Read requirements - Review USER_STORIES.md for the specific story
  2. Read related specs - Check API_SPECIFICATION.md, DATABASE_SCHEMA.sql, PROCESS_FLOWS.md
  3. Verify understanding - Confirm scope with Don if unclear
  4. Plan implementation - Identify files to create/modify
  5. Implement - Write code following ARCHITECTURE.md patterns
  6. Test - Verify against requirements
  7. Update docs - Update ARCHITECTURE.md if patterns change

Next Phase: Client Intake Implementation

Sequence 1: Client Intake (INT-001 to INT-006)

Story Description Priority
INT-001 New client registration portal P0
INT-002 Identity verification (Persona) P0
INT-003 Conflict check automation P1
INT-004 Engagement letter generation P0
INT-005 E-signature integration (Google) P0
INT-006 Initial document request P1

Prerequisites for Client Intake

  • Review INT-001 through INT-006 in USER_STORIES.md
  • Review client-related endpoints in API_SPECIFICATION.md
  • Create domain models (src/domain/client.py, etc.)
  • Create repository layer (src/repositories/client_repository.py)
  • Set up FastAPI routes (src/api/routes/clients.py)
  • Create basic API schemas (src/api/schemas/client_schemas.py)

Implementation Order (From USER_STORIES.md)

Sequence Name Stories Focus
1 Client Intake INT-001 to INT-006 Registration, identity verification, engagement
2 Document Collection INT-007 to INT-012 Upload, classification, extraction
3 Return Preparation INT-013 to INT-018 AI review, calculations, forms
... ... ... ...

Cloud Migration Queue

Items to address when moving from local dev to AWS:

Item Description Priority
Aurora setup Create Aurora PostgreSQL cluster P0
S3 buckets Create document storage buckets P0
Secrets Manager Move credentials from .env P0
IAM roles Service-specific roles P0
VPC configuration Network isolation P0
CI/CD pipeline GitHub Actions for deployment P1
Airflow EC2 Set up orchestration server P1

Pending Client Confirmations

See docs/CLIENT_FOLLOWUP.md for items requiring client input:

  • ASM-001: UltraTax handles all e-filing (assumed yes)
  • VND-001: SurePrep sandbox access
  • VND-002: SmartVault API credentials

To Resume Development

  1. Activate virtual environment: 

    source .venv/bin/activate

  2. Start database: 

    docker compose up -d

  3. Load environment: 

    source .env

  4. Continue with client intake domain models or API routes

Key Documents Reference

Document Purpose
ARCHITECTURE.md System design, code patterns
USER_STORIES.md 82 prioritized implementation stories
DATABASE_SCHEMA.sql Complete PostgreSQL schema
CLIENT_FOLLOWUP.md Pending client confirmations
backlog.md Technical debt and priorities
config.yaml Centralized configuration (Python + Java)