How OpenAI automated accounting with AI

Play video7:43

How do you build for the fastest growing company in history?

When Sowmya joined OpenAI in March 2023, ChatGPT had just become the fastest application to reach 100M users, creating unprecedented scaling challenges for the accounting team. With Excel breaking under the weight of their data, they needed to completely reimagine their financial operations to support growth to over 400 million weekly users.

The revolutionary solution was putting data at the center of finance rather than treating it as a downstream reporting function. Sowmya traded a revenue accounting headcount for a finance data engineer and built a system in which raw financial data flowed into Databricks (their data warehouse), transformed with SQL queries and Python scripts, and then went to NetSuite to feed real-time dashboards with automatic reconciliations. This infrastructure made accounting processes like revenue recognition and compute cost tracking completely "touchless."

Creating this system required both technical innovation and talent development. For compute costs that exceeded Excel's capacity, they implemented direct data feeds from vendors into their warehouse with automated Python scripts to update dashboards and create journal entries. When they couldn't find accountants with SQL skills, they hired for finance expertise and used ChatGPT to help team members develop technical capabilities over time.

Rather than mandating AI adoption, OpenAI fostered innovation through "finance hackathons" during team offsites. These low-pressure sessions allowed employees to experiment with ChatGPT for automating manual tasks, starting small with uploading CSVs to ChatGPT and eventually progressing to generating code. This bottom-up approach encouraged staff to identify their own pain points while still aligning with performance goals, creating a culture where finance professionals became comfortable leveraging AI tools for massive efficiency gains.