CrowdStrike to Acquire Fiddler AI for Agent Observability

Krishna Gade once helped explain why Facebook put a post in front of a person. At Fiddler AI, the explanation has had to follow software that can take actions, giving the record of a decision a longer and more consequential journey.

CrowdStrike has signed a definitive agreement to acquire Fiddler, announced November 3, 2026, at Fal.Con Europe in Barcelona. The proposed deal would bring AI agent evaluation, behavioral monitoring and explainability into the Falcon security platform as Falcon Observe. CrowdStrike targets general availability in Q4 FY27, putting the agent's decision history closer to the teams responsible for investigating enterprise risk.

The buying decision reaches across the organization. Engineers assessing whether an agent completed its task, security teams examining what it touched, and governance staff documenting the outcome have related questions. CrowdStrike is buying a way to connect the evidence those groups need, with integration and closing still ahead.

What CrowdStrike has agreed to buy

The official transaction announcement describes consideration predominantly in cash, with a portion delivered as stock subject to vesting conditions. CrowdStrike expects the acquisition to close during its fourth quarter of FY27, subject to customary closing conditions. The announcement gives no total purchase price, exact cash and stock allocation, or specific closing day.

Fiddler's contribution is a record of agent behavior across testing and production. CrowdStrike says the planned Falcon Observe offering will evaluate accuracy, task completion and policy compliance before deployment, then help trace decisions and unintended behavior once agents are operating. Those are announced product intentions, with general availability targeted for the same fiscal quarter as the expected close.

There is a practical reason for a security platform to want that material. An identity record can show which credential an agent used, while a behavioral trace can help explain how it arrived at an action. Connecting those records could let an investigator move between the assignment, the decision and its consequences without handing the question back to another team each time. That is the integration's operating proposition, rather than a measured customer outcome already established by the agreement.

Fiddler's founders have been following the explanation

Gade, Fiddler's CEO and co-founder, connects the company's origins to his work explaining Facebook ranking decisions in his account of the acquisition. He and Amit Paka founded Fiddler in 2018; Paka is its co-founder and COO. The Palo Alto company's work expanded through explainability, monitoring, evaluations and guardrails as the systems its customers used gained more responsibility.

That progression matters because the object being examined has changed. A model can produce an answer; an agent can use an answer to choose a tool, alter a resource or continue through a sequence of tasks. An enterprise investigating the result needs context from the sequence, including the instruction that made the action appropriate or inappropriate.

Fiddler's current control-plane offering describes evaluation, monitoring, policy enforcement and governance across first-party, third-party and coding agents. The vendor also describes tracking usage and costs alongside operational behavior. These capabilities put engineering quality, organizational policy and economic oversight in the same purchasing conversation, although a vendor's product description is no guarantee that a particular deployment will satisfy every requirement.

The founders' experience therefore has a distribution question attached to it. Work that began with explaining model behavior is proposed to sit inside a platform already selling enterprise security. How customers connect their own agents and processes will determine how useful that route becomes.

Venture funding built the business before the sale agreement

Fiddler announced a $30M Series C led by RPS Ventures on January 27, 2026, bringing total funding to $100M by the company's account. Participants included Lightspeed Venture Partners, Lux Capital and Insight Partners, alongside corporate and strategic investors. Those figures describe financing history, separate from the undisclosed official purchase price.

The earlier rounds show how long the monitoring problem has been funded. An Insight Partners-led $32M Series B was announced in June 2021. In December 2024, Fiddler announced an $18.6M Series B extension and described the total Series B as $50M. The historical announcements should retain their own accounting rather than becoming a homemade acquisition valuation.

CrowdStrike brings a different commercial scale. Its second-quarter FY27 results, for the period ended July 31, 2026, reported revenue of $1.47B and annual recurring revenue of $5.84B. Neither metric establishes the eventual economics of the Fiddler acquisition, but both help explain the distribution opportunity available through the Falcon platform.

For Fiddler's investors, the agreement follows years of financing an independent enterprise infrastructure company. For customers, the immediately useful questions concern integration scope, commercial terms and continuity of the workflows they already run. The announcement leaves those customer-specific answers open.

Behavioral evidence moves closer to enterprise response

CrowdStrike president Michael Sentonas describes the planned connection to Falcon security telemetry across endpoints, identity, cloud and runtime environments. The significance is the relationship between records: what an agent decided can be examined alongside what happened in the enterprise. That gives the proposed product a place in an investigation as well as an evaluation process.

CrowdStrike's separate Falcon Guardian announcement addresses controls for agent activity at runtime. Fiddler's proposed place beside that work makes explanation commercially relevant to the people expected to respond. A company expanding an agent's responsibilities also has to decide who can review its behavior and intervene when the assignment goes wrong.

For enterprise buyers, that decision can cross engineering, security and governance budgets. A combined platform could make evidence easier to share, while requiring customers to examine how it fits existing agents, frameworks and approval processes. Buyers can bring that review alongside AI spending and measurement, another part of the enterprise adoption decision.

Gade and Paka's work is now headed toward a security team's operating environment, subject to the deal closing and the product arriving. The next useful conversation for a customer can begin with an actual agent workflow: which decisions need to be recorded, which team receives the warning, and who has the authority to change what the agent does next.