Lightfield Raises $47M For AI-Native CRM Built Around Agents

Lightfield has raised $47 million to expand an artificial intelligence-native customer relationship management platform built for enterprise AI agents, SiliconANGLE reported.

The Series A financing gives the startup a sharper opening against entrenched CRM vendors such as Salesforce and HubSpot.

Andreessen Horowitz led the round.

Additional backers named in the round were Alumni Ventures, Audacious, Maverick Capital, Greylock, Coatue and Lightspeed Venture Partners.

The company, formally Magical Tome Inc, is positioning the fundraise around a specific systems problem rather than another layer of AI features.

Its premise is that more companies will rely on autonomous agents to handle parts of sales and customer work, but the data inside older CRM systems was organized for people filling in fields, notes and closing dates.

Agents that inherit that incomplete record can produce unreliable outputs because they cannot easily reconstruct what changed across an account or deal.

That architecture argument is central to the funding story.

A conventional CRM can hold notes from prior customer calls, a target close date and fields filled by a representative, but those fragments do not automatically explain the current state of a relationship.

Lightfield is betting that enterprises will need a cleaner operational record before they can trust agents with follow-up, forecasting or account work.

Lightfield’s product response is to make the CRM itself the agent-readable system of record.

The platform captures customer interactions across channels including email, calendars, Slack and LinkedIn, then structures those records so software agents and human sales teams work from the same picture.

Instead of waiting for a sales representative to summarize a call after the fact, the system is designed to join each customer touch point and refresh the underlying record as work happens.

Chief Executive Keth Peiris framed the reliability issue as a data problem, not simply a model problem.

His argument is that agents fail when the underlying business record is incomplete, inaccurate or too loosely structured, and that Lightfield can turn every customer exchange into a shared model of account reality and next actions.

The startup points to four technical elements behind that pitch.

One component refreshes the record as new interactions arrive.

A second maps emails, meetings and messages to the relevant accounts and deals while tracking changes over time.

A monitored agent layer keeps automated work inside a software development kit and sandbox.

A fourth design choice leaves the system open through APIs, command-line access and Model Context Protocol support, giving customers a way to build their own automations above the platform.

The early customer count gives Lightfield more than a concept to bring to investors.

Since launching its platform in November, the company has signed up more than 5,000 customers, from early-stage startups to scaling enterprises.

Peiris also pointed to dozens of companies that replaced Salesforce with Lightfield, making the startup’s challenge to a dominant incumbent part of its commercial story rather than only its product narrative.

The deal still asks customers to accept a major shift in where sales work is recorded and acted on.

Andreessen Horowitz general partner Alex Rampell tied the investment to a platform-cycle change, saying that each major platform shift creates a new system of record.

His closing comparison was blunt: Salesforce set that role for cloud software, while Lightfield is trying to define it for the agent era.