AI & Emerging Interfaces

WhiteMatter Raises $20M to Police AI Agents With AI Agents

WhiteMatter raised $20M for guardian AI agents that supervise enterprise agents. It already manages 2.25 million agents and 116 billion interactions.

  • AuthorSophie Lindqvist
  • ChannelAI & Emerging Interfaces
  • Filed
  • Read3 min
This startup raised $20 million to keep AI agents in line with more AI agents. Read the pitch deck.
This startup raised $20 million to keep AI agents in line with more AI agents. Read the pitch deck.Pixelkult / pixabay
  • WhiteMatter raised a $20 million Series A led by Blumberg Capital, with Crew Capital and Sprout & Oak participating.
  • The company reports 2.25 million agents under management and more than 116 billion AI interactions overseen.
  • Founded in 2024 as QuilrAI, it deploys one guardian agent per enterprise AI agent to check every file access, decision, and action.
  • The 80-person team is expected to grow by 50% within 12 months.
  • Customers include Fortune 200 and Fortune 500 firms such as Alter Domus.

WhiteMatter has raised a $20 million Series A to stop rogue AI agents — by deploying one supervisory AI agent for every agent an enterprise runs. The company says it already has 2.25 million agents under management and has overseen more than 116 billion AI interactions to date.

Founded in 2024 and formerly known as QuilrAI, WhiteMatter builds "guardian agents" that oversee what a company's existing AI agents can access, do, and share. The mechanism is specific: before an agent is allowed to read a file, use a piece of software, make a decision, or trigger an action, the guardian checks whether that action is permitted. The guardian then monitors the outcome. When an enterprise creates a new agent, a new guardian agent spawns automatically — the oversight layer scales in lockstep with the workload it watches.

What problem does this solve for practitioners?

The premise is that enterprise security stacks were designed around human actors, and agents break that assumption. "Our existing stack wasn't built for agents being the main actor in the story," Vidit Arora, the startup's 35-year-old founder and CEO, told Business Insider. "Organizations are going now from humans being the core actor to agents being the core actor, and our entire tool stack wasn't built for that."

Arora argues the current wave of agent-related security incidents is only beginning. We are "not even at the tip of the iceberg" when it comes to agent attacks and compromises, he said, both from external and internal agents. The context gives his claim weight: enterprises are granting non-technical employees access to agents that complete tasks autonomously, and a recent string of agent-powered security incidents has pushed agentic security up the agenda across the cybersecurity market.

How does the business model work?

The guardian agents do not consume customers' tokens. They run on WhiteMatter's own self-hosted AI model and GPUs. Pricing is a flat rate based on what the company calls a technical version of AUM: agents under management.

Growth, Arora said, is organic — companies keep deploying more agents, and each new agent needs its own guardian. Current customers include Fortune 200 and Fortune 500 companies, among them the asset management firm Alter Domus, plus firms in highly regulated industries such as healthcare.

Who funded it, and who are the rivals?

Blumberg Capital led the $20 million round. Existing investors Crew Capital and Sprout & Oak also participated. Arora plans to spend a large portion of the money on go-to-market and sales hires, and expects his roughly 80-person team to grow by 50% over the next 12 months. He also intends to invest in the company's research team.

WhiteMatter is not alone in the category. Its own website names Prompt Security, Lakera, and Protect AI as competitors. The company's name is a nod to the brain's connective tissue — the layer that carries signals between the parts that think.

The pitch deck the company used to raise the round lays out the same story investors bought: agents frequently take actions they are not meant to, guardian agents oversee every action an agent takes across its full lifecycle, and the market pays by the agent. If agent deployments keep compounding at their current pace, the bet is that oversight capacity — not agent capability — becomes the limiting factor in enterprise adoption.

Source: feeds.businessinsider.com

Share this article:

More from Sophie Lindqvist

Sophie Lindqvist

Show full bio

Staff writer covering people, launches and day-to-day practice in Product and interaction design, design systems at Interface Wire.

84 articles

Related articles

  1. OpenAI's 'Ladder' Framework Maps the Rise of Agent Swarms

  2. Agentic AI Systems Turn Fragile as Non-Engineers Build Big Balls of Mud

  3. CAISSI moves toward White House oversight without a permanent leader

  4. Zui Chen on Designing Interfaces When AI Takes the Wheel

« Previous articleNext article »