Monid raises $7.7 million to plug AI agents into pay-per-use tools

Monid, a San Francisco startup building what amounts to a toll booth for AI agents, has raised $7.7 million. The goal is to let agents tap pay-per-use tools and APIs through one platform instead of a drawer full of vendor accounts.

The round lands roughly a month after the company announced its pre-seed. That is a quick turnaround, even by the standards of an AI funding market that rarely pauses for breath.

What Monid actually does

The company describes its product as an agent-native tool router and marketplace. The comparison it invites is OpenRouter, which does something similar for large language models, except Monid focuses on the tools agents call rather than the models doing the thinking.

The workflow breaks into three steps: discover, inspect and run. An agent can search the catalog, look at what a tool does and how it compares to alternatives, then execute the call.

Developers integrate once and fund a prepaid balance. Every tool call draws from that balance, so there is no separate subscription for each provider and no stack of invoices at month’s end.

Pricing starts at approximately $0.0013 per call. Monid adds a 10% platform fee on top of provider rates, which is how the business makes money.

Two details stand out for developers who have been burned before. Failed calls are billed at zero, and new accounts get $1 in free credit to kick the tires.

The numbers behind the raise

Monid launched in April 2026 with approximately 200 tools. By September 2026, its catalog had grown to somewhere between 1,700 and 2,500+ endpoints, sourced from 55 to 72+ providers.

On September 1, 2026, the company announced a $2.1 million pre-seed. Backers included 1984 Ventures and Llama Ventures.

At that point, Monid said it had processed over 4 million agent transactions.

By October 6, 2026, the additional round of approximately $7.7 million had been completed. The capital is intended to scale the marketplace and the infrastructure underneath it.

Monid has also open-sourced its engine, which is built on Deno 2 and TypeScript, under the MIT license.

Why the API tax matters

Monid is pitching itself as a fix for what developers call the API tax. Every new capability an agent needs, whether that’s search, data enrichment or something else, usually means another account, another API key and another subscription.

Monid’s answer is per-call billing through a single balance. You pay for what the agent actually uses, not for a monthly plan that sits idle most of the time.

The model also changes who makes the decision. In Monid’s design, the agent itself can compare tools and pick one at runtime, rather than a developer hardwiring a single vendor months in advance.

What this means for the agent economy

That 10% cut is the business model, and its viability depends on volume. The 4 million transactions reported in September offer an early signal, though a per-call fee on fractions of a cent requires enormous scale to add up.

The open-source engine is a strategic hedge. Releasing the core under MIT can build developer trust and community contributions, though it also means anyone can study how the plumbing works.

Monid’s growth from about 200 tools to over 1,700 suggests providers are willing to participate. Billing failed calls at zero removes a common frustration with usage-based APIs, where developers pay for errors they did not cause.

A $2.1 million pre-seed followed within weeks by a $7.7 million round suggests backers see the agent tooling layer as a land grab worth entering early.