Lightfield raises $47M to build customer records for agents

A financing announcement after a product pivot

Lightfield announced a $47 million Series A on September 9, 2026. Andreessen Horowitz confirmed that it led the round. The transaction provides a concrete starting point for examining customer relationship management software designed around agent access, rather than a reason to assume the product has already displaced incumbent systems.

According to Andreessen Horowitz, Lightfield founders Keith Peiris and Henri Liriani previously built Tome. The investor said Tome had reached 25 million users before the founders pivoted. The history matters: this is the announced Lightfield Series A, not a claim that the founders or underlying legal entity had never raised capital before.

What the product and signup claims actually establish

Lightfield said more than 5,000 companies had signed up since its November 2025 launch. That is a company-reported signup count. It does not establish paid customers, completed migrations from another customer system, or continued usage after registration.

The company says its records can be read and written through API, MCP and CLI. These are vendor-described access methods. A buyer should ask for a demonstration of the actions permitted in a proposed workflow rather than infer the quality, permissions or completeness of the underlying data.

The commercial distinction is central. The supplied evidence supports a signup statement and a product capability statement, each attributed to the company. Combining them does not produce evidence that every registered company has entrusted its customer operations to autonomous agents.

The company and the funding participants

SiliconANGLE identified the legal company behind Lightfield as Magical Tome Inc.. That helps connect the current product to its corporate history without pretending that the new product announcement erased the earlier business.

SiliconANGLE also listed Lightspeed Venture Partners, Coatue, Greylock, Maverick Capital, Audacious and Alumni Ventures as participants. Investor participation establishes involvement in the transaction, not an independent audit of customer adoption. The lead investor’s memo likewise represents an investor’s account, not a separate assessment of paid conversion.

The supplied claims do not disclose a valuation, current revenue or a paid conversion rate. None should be reconstructed from the fundraising amount, the earlier user count or the new signup count.

A decision framework for Algerian B2B founders

For an Algerian business-to-business software founder, the useful connection is an indirect product and data question. Which customer records would a proposed agent need to read? Which should it be allowed to change? Who would check a change before it affects a commercial relationship? These are design questions, not claims about a verified local installation.

A buyer could test a bounded set of records, establish a human owner for corrections and request an explicit account of permissions. A founder could ask whether the product solves a customer-data problem sufficiently important to sustain payment. The signup figure cannot substitute for either exercise.

The pivot also invites a disciplined distinction between reach and commercial proof. An earlier product’s users and a new product’s registrations describe different audiences and measures. The financing supports a new direction, but its success should be judged through evidence appropriate to that direction rather than through inherited headline numbers.

Frequently Asked Questions

Sources & Further Reading

• Lightfield (via PR Newswire)
• Andreessen Horowitz
• SiliconANGLE