The fight to tame spreadsheets with AI continues. A new company called Meridian has emerged from stealth with a more comprehensive approach to agentic financial modeling, built around an IDE-like workspace, and it has secured significant funding to build it. This week, the company announced seventeen million dollars in seed funding at a one hundred million dollar post-money valuation.
The company’s goal is to make financial modeling and spreadsheets far more predictable and auditable. The CEO and co-founder, John Ling, explained the mission is to take a process that traditionally took several hours and condense it down to about ten minutes.
This funding round was led by Andressen Horowitz and the General Partnership. It also saw participation from QED Investors, FPV Ventures, and Litquidity Ventures. Meridian is already working with teams at Decagon and OffDeal, and signed five million dollars in contracts in December alone.
Excel agents have been a popular focus for AI startups, partly due to the high cost of human-led financial analysis. However, where previous agents were built directly into Excel, Meridian operates as a standalone workspace. This IDE-style approach allows the app to integrate various data sources and outside references more seamlessly, reducing friction.
Based in New York, the Meridian team includes alumni from AI firms like Scale AI and Anthropic, as well as financial veterans from firms like Goldman Sachs. According to Ling, the company’s biggest challenge is meeting the strict requirements of financial clients, which often clash with the non-deterministic nature of AI models.
He illustrated this by comparing software engineering to financial analysis. In software, ten engineers might produce ten different implementations for a new feature, which is acceptable. In finance, ten banking analysts asked for ten valuation models for a company would produce ten nearly identical workbooks. This demand for consistency is paramount.
Consequently, the Meridian team has focused intensely on making their AI outputs more auditable and deterministic while preserving the flexibility of large language models. The result is a blend of agentic AI and conventional tooling, designed to minimize the hallucinations that often slow enterprise AI deployments.
The ultimate aim is to remove the doubt from the LLM process. Users can see exactly how the logic flows and trace every assumption that goes into the model, understanding precisely where each element originates.