A San Francisco startup that promises to make decades-old desktop software talk to modern AI agents has closed a $3.8 million seed round, betting that the messiest corners of enterprise IT might be the most lucrative for automation.
Minicor announced Monday it raised the funding in a round led by Brad Gillespie at IA Ventures. The investor list reads like a who's who of early-stage tech: Y Combinator, Paul Graham, Pioneer Fund, Robinhood Ventures, Garage Capital, Rebel Fund, Blast, SuccessVP, K5 Global, 20VC, Phosphor Capital, and Metsu Capital all participated. Angel investors include Austen Allred, Henri Stern, Matteo Franceschetti, and Jeff Henriod, according to a company statement.
The pitch is straightforward, if technically daunting. Minicor builds an API layer that lets AI agents automate tasks inside legacy desktop applications that were never designed for modern integration. Think SAP interfaces from the 1990s. Electronic health records that still require manual clicking through dozens of screens. Warehouse management systems running on IBM AS/400 green screens.
These aren't edge cases. Minicor claims 70 percent of Fortune 500 companies still run on desktop apps with no API, though the figure lacks independent corroboration. A March 2025 report by Growth Acceleration Partners, a software consultancy, found 92 percent of organizations still depend on legacy technologies despite years of talk about modernization.
"Minicor lets AI use legacy desktop software like an API," said Faizaan Chishtie, the company's co-founder and CEO.
Capturing Workflows, Building Code
The platform works by capturing desktop workflows once, then running them as deterministic code while exposing a single API endpoint that AI products can call. The system handles on-premise Windows desktops, Citrix remote desktop sessions, and browser automations.
Use cases span industries where manual data entry remains stubbornly common. Doctors writing AI-generated clinical notes into electronic health records. Insurance processors moving claims through systems that predate the cloud. Data entry clerks feeding information into SAP, warehouse management platforms, or those stubborn AS/400 terminals that refuse to retire.
The company's homepage displayed 2.5 million executions run as of mid-October, a marketing metric not independently verified. Minicor also mentions SOC 2 Type II and HIPAA compliance in company blog posts, though the company did not provide links to third-party audit reports.
From Laminar to Liftoff
Chishtie and co-founder Saheed Akinbile, Minicor's CTO, started the company in 2024 under the name Laminar. They took a pre-seed round from NOMO Ventures before joining Y Combinator's Spring 2026 batch. The legal entity still operates as Laminar Run, Inc., according to the company's website footer.
"We've found PMF and just tripled our revenue since joining Y Combinator," Chishtie wrote on LinkedIn earlier this year. The claim suggests traction, though the company did not disclose revenue figures or customer counts.
Minicor lists 11 employees on LinkedIn across offices in San Francisco and Toronto, though the self-reported company size shows "2-10 employees."
Crowded Field, New Angle
The startup enters a market where traditional robotic process automation vendors are scrambling to adapt to agentic AI. UiPath rolled out "Agentic Automation" platform updates in late September. Automation Anywhere expanded support for AI agents and Model Context Protocol in its A360 v.42 release earlier this month. SS&C Blue Prism markets Robotic Desktop Automation products. Microsoft offers Power Automate Desktop for legacy UI automation.
Research firm Dabl Club recently mapped what it calls the "computer-use agent desktops" landscape, cataloging vendors enabling agents to drive desktops by screenshot and click. Forrester wrote in June that companies need orchestration and governance frameworks before agentic AI adoption can scale beyond pilots.
Minicor's angle is positioning itself as a bridge: combining deterministic desktop automation with AI to build what it calls "self-healing workflows," then wrapping the whole thing in an API that developers can call from any application. Whether that's enough differentiation in a market thick with automation tools remains to be seen.
For now, the company has runway and a thesis that legacy software, far from being a liability, might be the next frontier for AI automation. Sometimes the oldest problems attract the freshest capital.
