The combat to tame spreadsheets with AI isn’t over but. A brand new firm known as Meridian has emerged from stealth with a extra complete IDE-based method to agentic monetary modeling — and loads of funding to construct it. On Wednesday, the corporate introduced $17 million in seed funding at a $100 million post-money valuation.
“Our purpose is to make monetary modeling and spreadsheets far more predictable and auditable,” CEO and co-founder John Ling advised TechCrunch. “How will you take a course of that historically may need taken cool hours and condense it down into like 10 minutes?”
The spherical was led by Andressen Horowitz and the Common Partnership, with participation from QED Buyers, FPV Ventures, and Litquidity Ventures. The corporate says it’s at present working with groups at Decagon and OffDeal, and signed $5 million of contracts in December alone.
Excel brokers have been a preferred goal for AI startups, due partially to the excessive price of human-led monetary evaluation. However the place earlier Excel brokers like Shortcut AI constructed brokers into Excel, Meridian operates as a stand-alone workspace, extra akin to Cursor. This permits the app to function like an IDE, integrating knowledge sources and different exterior references that may in any other case create friction.
Based mostly in New York, the Meridian crew contains each alumni of AI companies like Scale AI and Anthropic in addition to monetary veterans from companies like Goldman Sachs.
As Ling describes it, Meridian’s largest problem is the strict necessities of economic shoppers, which regularly conflict with the non-deterministic nature of AI fashions.
“When you go to 10 completely different software program engineers at Google, and also you need to add some new function into an app, you’ll in all probability get like 10 fully completely different implementations. And that’s completely nice,” Ling says. “However when you go to 10 banking analysts at Goldman Sachs and also you ask for 10 valuation fashions for an organization, you’d in all probability get 10 virtually equivalent workbooks.”
Techcrunch occasion
Boston, MA
|
June 23, 2026
In consequence, the Meridian crew has carried out important work to make their outputs extra auditable and deterministic, whereas sustaining the flexibleness of LLM-based instruments. The result’s a combination of agentic AI and extra standard tooling, minimizing the hallucinations that decelerate many enterprise deployments.
“Our purpose is to essentially take away the doubt layer proper from the LLM course of,” Li says. “You understand precisely how the logic flows, and all of those assumptions or no matter that go into the mannequin, you may see precisely the place they’re coming from.”