Stigg 2.0 Decides What Every AI Request Is Allowed to Cost, in Under Five Milliseconds

The usage runtime for AI products: Stigg has rebuilt its platform to enforce credits, entitlements, and budgets on every request and to deploy into a customer's own cloud.

SAN FRANCISCO, June 30, 2026 /PRNewswire/ -- Stigg today released Stigg 2.0 at the AI Engineer World's Fair, a real-time layer that keeps AI usage under control by deciding what every customer, user, team, and AI agent is allowed to do at the moment they try to do it.

AI billing breaks at the request, not the invoice. Older systems assumed the monthly invoice settled the account. In AI, the decision that matters happens the moment a request arrives and the cost is incurred. A single API call can cost dollars, and one agent can spawn sub-agents firing dozens of parallel calls against a shared credit pool, leaving a system single-digit milliseconds to authorize the work or absorb the bill.

The largest AI companies reached this conclusion and built the infrastructure themselves. In February 2026, OpenAI described a "decision waterfall" that synchronously checks rate limits and verifies credits before returning a single outcome, then settles the debit afterward. OpenAI wrote that it had evaluated third-party metering and billing platforms and found them suited to invoicing but unable to make that decision in real time. InfoQ reported the same shift, placing OpenAI's approach alongside Uber's move to an infrastructure-level rate limiter.

"We read that post and recognized our own architecture," said Anton Zagrebelny, Co-Founder and Chief Technology Officer of Stigg. "OpenAI can staff a team to build this. Most companies cannot. You can write a credit counter in a weekend. Two months later it handles a third of the edge cases, and the rest arrive as billing disputes, revenue leakage, and 3 AM pages."

The pricing shift behind it is broad. AI-native companies are moving off per-seat pricing toward credits and consumption, according to Bessemer Venture Partners, and the share of companies on the PricingSaaS 500 index using credit-based pricing rose 126% in 2025. The exposure lands at the account level. Stigg customers report a single power user exhausting an organization's entire AI credit allocation in one day.

"Enterprises will not sign six-figure AI contracts without governance," Zagrebelny said. "They want to know they can cap a runaway agent before it spends fifty thousand dollars on inference. That is an entitlements question, not a billing one."

Stigg 2.0 treats entitlements, the record of what a customer is actually allowed to use, as infrastructure that sits alongside billing rather than inside it. The release includes a rebuilt credits engine with real-time, zero-overdraft balances; a governance layer that evaluates budgets and limits across any dimension in under five milliseconds; and a metering pipeline that sustains more than one million events per second. A modular bring-your-own-cloud architecture runs any component, the metering pipeline included, inside a customer's own cloud, so the most sensitive usage data never leaves their infrastructure. A Model Context Protocol server lets AI agents check entitlements and report usage directly. Stigg has also acquired Received.ai for invoicing and contract management and added an Airwallex integration that gives early-stage startups free end-to-end billing.

"Metering is the highest-volume, most sensitive data an AI company holds, every call, every token, every customer's usage pattern," said Dor Sasson, Co-Founder and Chief Executive Officer of Stigg. "The industry default has been to ship all of it to a vendor's cloud and hope the security review passes. That model does not survive enterprise procurement or data-residency requirements. Putting the metering pipeline inside the customer's own cloud, enforced in real time, is not a deployment option. It is where the category is going."

Stigg 2.0 is available today, with a free tier that requires no sales call. Documentation and SDKs are at docs.stigg.io.

About Stigg

Stigg is a pricing and entitlements infrastructure company whose real-time layer sits between an application and its billing stack and decides what each customer, user, team, and agent is allowed to do. Founded by Dor Sasson and Anton Zagrebelny, Stigg works alongside existing billing systems to add the credits, entitlements, metering, and governance that AI products need. Stigg is SOC 1 Type 2 certified. Learn more at stigg.io.

Photo - https://mma.prnewswire.com/media/3002479/Stigg.jpg

Media Contact
Mor Levy Gaier, VP Marketing, [email protected] | stigg.io

SOURCE Stigg