13. Paystack Integration

Chapter 13 of 24 · 15 min

KEY INSIGHT

Payment integration is where operator-level rubber meets the road. With local models, you're not paying per-API-call, which fundamentally changes your revenue model—but you still need payment rails that don't compromise user data sovereignty. Paystack's API handles African market payments, but the integration pattern matters for local AI products. Your users expect their queries to never leave their infrastructure. If you route payment data through third-party systems with aggressive logging, you undermine the core value proposition. ```python # paystack_integration.py import hashlib import hmac import time class LocalAIPaystackIntegration: def __init__(self, secret_key: str): self.secret_key = secret_key self.api_base = "https://api.paystack.co" def create_generation_key(self, user_id: str, plan: str) -> dict: """ Generate a license key that's validated locally. Payment reference stored on user's own infrastructure. """ timestamp = int(time.time()) payload = f"{user_id}:{plan}:{timestamp}" signature = hmac.new( self.secret_key.encode(), payload.encode(), hashlib.sha256 ).hexdigest() return { "license_key": f"LM-{signature[:16].upper()}", "plan": plan, "issued_at": timestamp, "expires_at": timestamp + (86400 * 30), # 30-day validity "payment_ref": f"PS-{signature[16:32]}" # Reference for user's records } def verify_webhook(self, payload: bytes, signature: str) -> bool: """Verify webhook authenticity for local processing.""" expected = hmac.new( self.secret_key.encode(), payload, hashlib.sha256 ).hexdigest() return hmac.compare_digest(expected, signature) ``` The critical design decision: your local AI product doesn't need to store payment data centrally. Generate license keys locally after payment verification. The user's system maintains the authoritative record. For usage-based billing with local models, you track token consumption on-device and report aggregated usage monthly. This respects data locality—individual prompts never leave the user's infrastructure.

Local verification checkpoint

Run the smallest example from this chapter in a local workspace and record the package version, runtime, data path, and observed output. If the result depends on model size, vector count, CPU/GPU backend, or available memory, note that constraint beside the exercise so the lesson remains reproducible.

EXERCISE

Design a billing system where you never receive user prompt data. Sketch a flow where licensing is confirmed via webhook, but usage metrics are aggregated locally and reported as totals. Implement the license validation function that checks expiry without external calls.