AcquiredStartups
Acquisition · AI · United States

Beacon Software acquires AI red-teaming startup Haize Labs

Haize Labs, valued at $100 million in 2024 after a round led by General Catalyst, will stop being a standalone vendor and become the in-house AI research group for Beacon's 45 software companies.

ASAcquired deals deskNorth America17 September 2026 · 3 min read
Beacon Software acquires Haize Labs
Photo: King of Hearts / Wikimedia Commons, CC BY-SA 4.0

Beacon Software, a Toronto-based company that buys small software businesses and adds AI to their products, said on September 17, 2026 that it had acquired Haize Labs, a New York startup that tests and safeguards AI systems before and after they are deployed.

Terms were not disclosed. The Haize Labs team will form Beacon's new Applied AI Research Group. Co-founder and chief executive Leonard Tang joins Beacon as vice president of AI research, where he will lead the shared AI operating system that Beacon deploys across its portfolio companies. Co-founder Steve Li is also moving to Beacon, according to an analysis by BERI.

The announcement did not say what will happen to Haize Labs' existing enterprise customers or whether its platform will continue to be sold to outside clients.

What Haize Labs built

Haize Labs was founded in 2024 by Tang, Richard Liu and Steve Li, who met as undergraduates at Harvard, PitchBook reported. The company focused on preventing AI jailbreaks, using automated red-teaming to stress-test models and agents before deployment and to monitor them in production, with attack-generation software it called the haizing suite. It also sold guardrails, evaluation and observability tools, and did red-teaming and safeguards work with frontier AI labs.

VentureBeat reported in June 2024 that Anthropic was a client, PitchBook noted. Haize also released Verdict, an open-source library for using language models as judges of other models' output, under an MIT licence. Its hosted platform and attack-generation software were not open source, according to BERI.

“The Fortune 500 will have no shortage of AI. The harder problem is making AI useful for the businesses that power the rest of the economy, where most humans work and receive value.”Leonard Tang, Co-Founder and CEO, Haize Labs
Funding historyTotal raised n/a
2024Venture roundGeneral Catalystn/a

The founders and the funding

About seven months after it was founded, Haize Labs was valued at $100 million post-money in a competitive round led by General Catalyst, according to PitchBook, which cited two people familiar with the deal. PitchBook could not learn how much the company raised. It reported that Haize had received at least four term sheets valuing it between $30 million and $100 million, including one from Coatue.

Tang was 22 at the time of that round. Haize Labs did not disclose any later funding, its revenue or its headcount.

Why Beacon is buying

Beacon calls itself the first centralised AI holding company. It acquires software businesses that serve traditional industries and connects them to a common platform for engineering, AI, product and sales. Its portfolio now includes 45 software companies with customers in recreation, utilities, education, government, manufacturing and other sectors.

Former Instacart executive Nilam Ganenthiran founded Beacon in January 2024. In June 2026 the company raised a $225 million Series C led by General Catalyst and HarbourVest, with Intrepid Growth Partners also investing, The Logic reported. That followed a $250 million Series B in November 2025. At the time of the Series C, Beacon owned more than 30 businesses and was adding about one a week. General Catalyst therefore backed both the buyer and the startup it is acquiring.

Ganenthiran said Haize Labs brings engineers who know how to find where AI fails and how to fix it. Beacon listed four priorities for the new research group: building reusable AI infrastructure that works across the portfolio while keeping each company's local context; turning industry knowledge into evaluation and feedback systems; continuous testing, red-teaming and monitoring of AI agents; and building AI products that cut administrative work and improve customer service.

From vendor to internal team

The deal turns a company that sold AI safety testing to large technology firms into an internal team serving smaller, traditional businesses. Tang framed the move in those terms, saying large enterprises will not lack for AI and that the harder problem is making it useful for the rest of the economy.

BERI noted that the announcement makes no commitment on continuity for Haize's outside customers and advised them to seek written confirmation of support dates, contract ownership and notice periods.

Sources