Tollgate is a real-time margin observability platform designed for companies that deploy AI agents. Its primary function is to provide live gross margin tracking per customer, per agent, and per run by joining what a company charges with what each AI run actually costs. The tool addresses the gap between billing data from payment processors and cost data from LLM providers, enabling organizations to see profitability at a granular level rather than relying on blended metrics that can hide unprofitable customers or workflows. Key features include per-customer and per-agent profitability dashboards, real-time AI cost accounting, and margin alerts that notify users when a customer or agent becomes unprofitable. The platform supports tracking of reasoning tokens, which can cause individual runs to cost significantly more than normal ones, and provides CSV export of costs, logs, and margins. Tollgate offers tiered plans: a free tier with 10,000 ingest events per month, 3 tracked customers, and 1 workspace seat; a growth tier with 500,000 events, 50 customers, 5 seats, email and webhook alerts, and CSV export; and an enterprise tier with 5 million events, unlimited customers and seats, role-based access control, SSO/OIDC integration (Google, Okta, Azure), and queryable compliance audit logs. Typical use cases include identifying which customers or agents generate healthy margins versus those that consume more AI spend than they produce in revenue, enabling informed pricing decisions, and scaling profitability analysis beyond manual spreadsheet reconciliation. The workflow involves ingesting event data from the company’s AI agent runs, linking that data to customer billing information, and surfacing margin insights through the dashboard or alerts. Tollgate is particularly relevant for companies that have grown beyond a handful of customers, where manual tracking becomes error prone and blended metrics can mask significant losses.