General Analysis Raises $10M Seed to Secure Agentic AI

General Analysis announced a $10 million seed round led by Altos Ventures, with participation from 645 Ventures, Menlo Ventures, Y Combinator, and additional strategic investors and angels, per a Business Wire press release published April 29, 2026. The company says it was founded in 2025 by Rez Havaei, Maximilian Li, and Rex Liu; Business Wire and FinSMEs describe the founders as former researchers at NVIDIA, Cohere, DeepMind, Harvard, and Caltech. Business Wire reports that General Analysis is already working with enterprise customers in support and finance whose products and workflows reach hundreds of millions of users. FinSMEs reports the company intends to use the funds to expand operations and development efforts.

Per the Business Wire release, General Analysis ran adversarial evaluations in March that targeted live customer-service AI agents. Business Wire states the adversarial agent was able to induce roughly 50 live agents to offer fabricated perks totaling more than $10 million in simulated value across targets, and that only 5 out of 55 bots refused during the tests.

According to Axios, co-founder Rez Havaei told Axios Pro the company aims to hit $2 million in revenue in the next 12 to 18 months and then will look to raise a Series A.

Industry observers increasingly treat security for agentic AI as a distinct technical domain rather than an extension of traditional application security. Companies building and deploying autonomous agents face failure modes that emerge from model behavior, stateful interactions, and API integrations, rather than from exploitable code paths alone. Observed patterns in comparable projects show that adversarial evaluation frameworks and red-team-style stress tests are effective early experiments for uncovering complex failure modes such as data exfiltration, logic-bypass, or reward-manipulation.

Investors adding capital to startups focused on agent safety and tooling is consistent with a broader financing trend: as enterprises move toward agentic automation in support, finance, and other high-volume workflows, demand for specialized defensive tooling and governance increases. Reporting on this round places General Analysis alongside other small vendors that combine adversarial testing, monitoring, and defensive controls for LLM-driven agents. Observed patterns in similar market segments suggest early customers are large enterprises that prioritize controlled pilot deployments and compliance-ready audit trails.

• •Product and go-to-market: monitor whether General Analysis publishes reproducible tooling, open benchmarks, or case studies from the described adversarial tests; the press release claims active enterprise engagements.
• •Customer traction and revenue: Axios reports a target of $2 million ARR in 12 to 18 months; actual contract sizes and renewal metrics will indicate whether the model is enterprise-scalable.
• •Technical disclosures: look for public writeups or reproducible artifacts from the March adversarial evaluations to validate scope and exploit classes described in the press materials.

Editorial analysis: Organizations deploying agentic systems should treat adversarial evaluation and scenario-based stress testing as part of a layered security program. The incidents described in the press materials illustrate how agentic failures can produce high-value downstream effects that are not visible from static code review alone. Teams building or securing agents will benefit from integrating red-team findings into CI/CD testing, observability pipelines, and incident response playbooks.