# p90

> P90 is AI product pricing intelligence. It monitors cost variance, models fully loaded serving cost, and supports human-approved pricing decisions.

## Product
- [About](https://p90.tech/about)
- [Documentation](https://p90.tech/docs)
- [Connect a product](https://p90.tech/dashboard/connect)

## Guides
- [AI pricing models that survive real usage](https://p90.tech/guides/ai-pricing-models): How to choose subscription, usage-based, hybrid, and outcome-based AI pricing models without losing control of gross margin as customer behavior changes.
- [How to calculate AI product gross margin](https://p90.tech/guides/ai-gross-margin): A clear method for calculating AI gross margin, finding the real cost to serve, and using margin data to make better product and pricing decisions.
- [AI cost monitoring: what to track beyond token spend](https://p90.tech/guides/ai-cost-monitoring): A practical AI cost monitoring framework for product, finance, and engineering teams managing model, retrieval, retry, and concurrency cost.
- [AI unit economics: a practical model for product leaders](https://p90.tech/guides/ai-unit-economics): How to build an AI unit economics model that connects a customer-facing value unit to direct cost, revenue, gross margin, and the operational choices behind them.
- [Usage-based pricing for AI products without confusing customers](https://p90.tech/guides/usage-based-pricing-ai): A practical framework for usage-based AI pricing that keeps invoices understandable, aligns a pricing metric with customer value, and protects margin at high usage.
- [How to price AI agents when the work is variable](https://p90.tech/guides/ai-agent-pricing): A practical guide to AI agent pricing for products where tool calls, retries, model routing, and task complexity make the cost of each outcome uneven.
- [RAG cost optimization: reduce retrieval spend without reducing answer quality](https://p90.tech/guides/rag-cost-optimization): How to approach RAG cost optimization by measuring retrieval quality, context growth, model usage, and gross-margin impact instead of simply cutting documents or tokens.
- [AI pricing experiments that produce useful evidence](https://p90.tech/guides/ai-pricing-experiments): How to run AI pricing experiments that test customer value, usage behavior, willingness to pay, and gross-margin resilience without creating a confused offer or unreliable conclusion.

## Use cases
- [AI SaaS pricing](https://p90.tech/use-cases/ai-saas-pricing): Set a defensible floor, design tier boundaries, and review the margin impact of real product behavior.
- [AI cost monitoring](https://p90.tech/use-cases/ai-cost-monitoring): Connect product telemetry, provider data, and revenue so cost drivers are visible before they become a margin issue.
- [Agentic AI pricing](https://p90.tech/use-cases/agentic-ai-pricing): Price variable agent workflows without hiding retries, tool calls, context growth, or long-running execution inside an arbitrary subscription.
- [RAG pricing](https://p90.tech/use-cases/rag-pricing): Understand how retrieval volume, context assembly, and document complexity change the cost of a RAG product before packaging it.
- [Enterprise AI deal economics](https://p90.tech/use-cases/enterprise-ai-deal-economics): Test workload, concurrency, SLA, and growth assumptions before committing to an enterprise contract.
- [Outcome-based AI pricing](https://p90.tech/use-cases/outcome-based-ai-pricing): Assess whether an AI product has enough evidence to price a measurable result rather than an internal technical meter.

## Integration
P90 SDK sends product telemetry. MCP provides workspace monitoring context and bounded, explicitly confirmed actions through workspace-scoped service API keys.