Lightfield closed a $47 million Series A led by Andreessen Horowitz, with Maverick Capital, Coatue, Greylock, Lightspeed, Audacious and Alumni Ventures joining, to build what the company calls a system of record designed for AI agents rather than for people typing into fields.
The pitch is specific: agents layered onto Salesforce or HubSpot inherit whatever a rep had time to enter by hand, incomplete fields, stale close dates, notes written from memory after a call, and produce work nobody can verify. Lightfield instead ingests every customer interaction, email, calendar, calls, Slack, LinkedIn, directly and structures it into a record built for agents to reason from, exposed through API, MCP, and CLI so any automation a company builds can read and write to it. More than 5,000 companies have signed up since the product launched in November 2025, some of them replacing Salesforce outright.
Keith Peiris, Lightfield’s cofounder and CEO, framed the failure point plainly. “Agents don’t fail because the models aren’t capable. They fail because the data they work with is incomplete, inaccurate, and missing the structure needed for comprehension,” Peiris said. Alex Rampell, general partner at Andreessen Horowitz, went further: “Salesforce defined it for the cloud era, and Lightfield is defining it for the agent era.”
The original insight is what the round says about where AI agent spend is actually concentrating in revenue tech right now. It is not going to companies promising a smarter agent on top of the CRM everyone already has. Investors backed a company betting the CRM itself is the broken layer, which is a harder, slower rebuild than another AI feature, and a much bigger bet if a16z is right that it holds.
Source: PR Newswire
Related: Owning the AI Stack Becomes RevOps’ New Ask | The AI Maturity Gap Has One Root Cause: CRM Data