Ramp Labs has emerged as an experimental hub inside Ramp, focused on applied Artificial Intelligence to transform how finance work gets done. The group is given no traditional roadmap or key performance indicators, and instead is tasked with exploring ambitious use cases that appear impossible and iterating rapidly on them. Early experiments included embedding Claude Code inside RollerCoaster Tycoon to prove that Artificial Intelligence agents can navigate complex interfaces, manage operations, and make real time financial decisions, illustrating how similar technology could be translated to real businesses. That exploratory culture set the stage for what became Ramp Sheets, an Artificial Intelligence powered spreadsheet editor designed specifically for finance professionals.
Ramp Sheets began as an attempt to automate Ramp’s own finance workflows, after the team found that there was a 95% chance that an internal user was in Excel at any random moment and that every major finance process depended on spreadsheets. Initial efforts to generate Python scripts failed because finance teams could not trust or verify unfamiliar code, so the team pivoted to putting Artificial Intelligence directly inside a traditional spreadsheet environment. After two months of intense iteration, the product launched and quickly gained traction without paid marketing or a public relations agency, signing up thousands of users and generating 2 million impressions in its first week. Usage has since grown into one of the most used Artificial Intelligence spreadsheet tools in finance, with teams at KKR, Thrive Capital, General Catalyst, Wharton, Stanford, and MIT adopting it for daily work.
The core concept of Ramp Sheets is a familiar spreadsheet interface augmented by an Artificial Intelligence agent that can build models, clean data, write formulas, search the web, and format outputs using natural language commands, while maintaining full transparency into every formula and cell adjustment. Finance professionals can request tasks such as “build a 13-week cash flow forecast for a 50-person SaaS company with $8M ARR” and get a board-ready model, or “benchmark our engineering compensation against San Francisco market rates” and receive a structured comparison with current salary data. Used cases now include leveraged buyout modeling in a fraction of the previous time, startup valuations for fundraising, and automated monthly close processes. The broader vision extends beyond spreadsheets into what Ramp describes as “thinking money,” where Artificial Intelligence agents track and control spend across a business. In October, Ramp’s Artificial Intelligence made 26,146,619 decisions across more than $10 billion in spend, preventing 511,157 out-of-policy transactions that saved $290,981,801, moving $5.5 million from idle cash to 4% investments, blocking a $49,000 Artificial Intelligence generated fake invoice, and even saving $113.34 on an individual employee’s travel booking, with the goal of freeing finance teams to focus on higher leverage strategic decisions.
Ramp positions this shift as the beginning of an era in which every dollar “thinks” by checking permissions, auditing itself, and judging whether it was well spent, reversing the drift toward bureaucratic sprawl that often follows growth. The company argues that this model restores economic productivity by automating countless small but objective decisions that humans frequently miss, citing that the median Ramp customer saves 5% while growing revenue 12% year over year and closes books in days instead of weeks. Internally, everyone from engineers to marketers reportedly uses Artificial Intelligence tools such as Claude Code, Cursor, and Ramp’s proprietary systems, which are described as having 10x’d shipping velocity. Ramp Labs continues to explore generative user interfaces that adapt to each user, video-based process mining that can derive process maps from Loom recordings, and reinforcement learning to make its agents faster and cheaper.
