> ## Documentation Index
> Fetch the complete documentation index at: https://docs.paygentic.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> Infrastructure for consumption-based monetization

Paygentic is infrastructure for consumption-based monetization. Built for software where costs scale with usage — LLMs, data processing, compute resources — not user seats.

## Why Paygentic Exists

Software economics changed. Fixed monthly subscriptions break when your costs vary per request. When running GPT-4 costs \$0.03 per thousand tokens, charging \$20/month regardless of usage creates unsustainable unit economics.

The technical challenges multiply:

**Metering at Scale**
Tracking millions of events per second — tokens processed, queries executed, bytes stored — without impacting application performance.

**Pricing Flexibility**
Supporting per-token, per-query, per-gigabyte, and hybrid models simultaneously across different customer segments, with both upfront and usage-based charges in the same plan.

**Payment Collection**
Collecting upfront fees before a subscription activates, then billing metered usage at period end — all through a single integration.

**Subscription Lifecycle**
Managing activations, renewals, upgrades, and cancellations with correct proration and invoicing at every transition.

## Architecture Philosophy

Paygentic unifies metering, pricing, invoicing, and payments into a single pipeline. You define what to measure, how to price it, and when to bill — Paygentic handles the rest.

This design eliminates entire categories of problems:

* No reconciliation between usage logs and billing records
* No separate integration points for metering vs invoicing vs payments
* No manual invoice generation or payment chasing
* No custom logic to gate activation on upfront payment

## Core Primitives

**Products & Metrics**
Define services and measurable units. An LLM service tracks input tokens, output tokens, and fine-tuning hours as distinct billable metrics.

**Plans & Prices**
Package pricing strategies. Standard rates, dynamic pricing, or percentage fees — each with a payment term (`in_advance` for upfront charges, `in_arrears` for metered usage). Plans bundle prices into subscribable offerings with a billing interval and currency.

**Subscriptions & Invoices**
Connect customers to plans. When a subscription includes upfront charges, Paygentic generates an initial invoice that must be paid before the subscription activates. Subsequent invoices are generated automatically at each billing period.

**Meter Events**
Report consumption as it occurs. Each event carries a customer ID, metric, and quantity. Events are fire-and-forget — always `202 Accepted` — and Paygentic aggregates them for invoicing at period end.

**Features & Entitlements**
Grant access to capabilities through your pricing. Attach entitlements to prices with quota limits, reset periods, and rollover behavior. When a customer subscribes, entitlements are provisioned automatically.

## Implementation Patterns

**AI Services**

* Input/output token differentiation
* Model-specific pricing tiers
* Platform access fees (in advance) plus token usage (in arrears)
* Fine-tuning compute billing

**Data Platforms**

* Storage per GB-month
* Query compute seconds
* Egress bandwidth charges
* Reserved vs on-demand pricing

**Developer Tools**

* Build minutes
* Deployment slots
* Bandwidth consumption
* Concurrent execution limits

## Technical Integration

Single API surface. Define your monetization model programmatically:

1. Create products and billable metrics
2. Configure prices and plans
3. Onboard customers and create subscriptions
4. Report meter events
5. Invoices are generated and collected automatically

No payment gateway complexity. No manual reconciliation.

## Design Principles

**Real-time by default.** Meter events are ingested and aggregated in real time. Invoicing and settlement follow automatically.

**Precision matters.** Atomic unit accounting enables sub-cent transactions at scale.

**Idempotent everything.** Network failures don't create duplicate charges or lost events.

**Geographic distribution.** Edge processing for low-latency event ingestion globally.

## Getting Started

Follow the [Quickstart](/getting-started/quickstart) for a hands-on walkthrough. Then explore [Pricing](/platform/pricing/overview) to model your monetization strategy.