Pay-i Inc., a startup that helps companies track the financial footprint of their artificial intelligence workloads, launched today with $4.9 million in funding.

Fuse Partners and Tola Capital co-led the seed investment. They were joined by Firestreak, Pear VC, Gaia Capital and multiple angel investors.

Pear-i is led by Chief Executive Officer David Tepper, a former principal project management architect at Microsoft Corp.’s Azure business. He launched the company last year with longtime startup executive Erik Winters and former Windows architect Doron Holan. Winters is Pay-i’s chief operating officer and Holan is chief technology officer.

The company provides a platform that enables enterprises to collect financial information about their AI applications. Pay-i can track the amount of revenue that a service generates, the cost of operating it and related metrics. The software can also forecast future financial performance.

Many AI applications are powered by cloud-hosted large language models that are billed based on usage. OpenAI, for example, charges customers based on the amount of input and output data generated by their workloads. Other major AI providers such as Anthropic PBC have adopted a similar business model.

Pay-i visualizes the expenses associated with AI model usage in a bar chart. Each bar describes the AI costs incurred during a single day. According to Pay-i, its platform tracks how daily spending changes over time and breaks down the bill by application.

The software is also capable of monitoring an AI workload’s profit margins. To boost financial performance, a company can use a built-in A/B testing tool to pilot several different versions of the workload side-by-side. The information collected by the testing tool makes it possible to identify the version with the biggest revenue contribution.

Developers building a new AI application can use Pay-i to forecast the workload’s inference costs before it launches. That makes it easier to avoid budget overruns. If Pay-i determines that the AI model used by an application may make it difficult to meet profit margin targets, developers can switch to a lower-cost model before launch.

The company says that its platform also lends itself to tracking non-financial metrics. Pay-i collects usage statistics such as an application’s daily active users, the number of requests that it processes per hour and the amount of data in those requests.

The platform doubles as a troubleshooting tool. When an application experiences technical issues, Pay-i can highlight the prompt or AI agent step that caused the malfunction. It also provides higher-level metrics such as the percentage of user prompts that lead to errors.

“The C-suite doesn’t need another usage chart – they need proof and a forecast,” Tepper said. “Pay-i pinpoints which gen AI use cases create net-new value today, quantifies that value in dollars or hours, and predicts how it will compound tomorrow.”

Pay-i will use the proceeds from its seed round to drive product adoption and add more features. 

Image: Pay-i