Ent raises $100M seed to bring intent-aware prevention to endpoints

Ent, a San Francisco startup building intent‑aware endpoint security, emerged from stealth on June 16, 2026 with a $100 million seed round — a jumpstart few early-stage companies receive.

The financing was led by Decibel, with participation from Sequoia affiliates and other investors including Crosspoint, Craft Ventures, Shield Capital, Felicis and In‑Q‑Tel, the strategic investor for U.S. intelligence and defense programs.

Ent was founded by Elias “Lou” Manousos and Brandon Dixon, veterans of RiskIQ and contributors to Microsoft’s Security Copilot work, and the company says the team’s prior platform experience informed its prevention-first strategy.

The startup’s core pitch is simple: run small, intention‑aware models on the device to infer what a user or an agent means to do and stop risky actions before they complete. Ent calls this “intent-aware Workspace Security,” framing prevention — not after‑the‑fact detection — as the primary objective.

Technically, Ent says it builds a continuous record of work on the endpoint, uses on‑device models to infer intent at decision time, and intervenes at the moment of risk rather than after an event has occurred. The company positions that capability as crucial in an era of autonomous agents and model‑driven attacks.

Ent’s announcement says the new capital will accelerate hiring across engineering and go‑to‑market, and fund product work in AI governance, integrations and endpoint intelligence. The company intends to scale pilots and product development with the fresh funding.

The fundraising underscores a wider industry shift: several vendors and buyers now argue AI compresses the time from compromise to impact, meaning prevention must work at machine speed if defenders are to keep up. Ent and its backers frame the effort as reversing a decade where the market prioritized detection and response.

Investors and reporters have noted the round’s size: a $100 million seed is unusual and signals heightened investor interest in tooling for the ‘agentic’ enterprise, where employees and automated assistants both make security‑relevant decisions. That interest has pushed a handful of startups and incumbents to experiment with agent‑aware controls.

Ent’s on‑device model approach raises familiar tradeoffs. Enterprises will watch for how the product balances privacy, latency and false positives; on‑device inference can limit data exfiltration risk, but it must also avoid disrupting legitimate work or creating new operational overhead. Ent’s public materials highlight work with regulated customers as a design priority.

Reports around the launch say Ent plans to pursue Global 2000 customers across finance, hospitality and defense, sectors where a single human decision at the wrong moment can trigger outsized damage. Early go‑to‑market signals point at pilots and integration work with large enterprises.

What to watch next: whether Ent can demonstrate low‑friction prevention at scale, how it measures intent accuracy and false positive rates, and which customers validate prevention‑first controls in production. The company’s rapid capital raise gives it runway, but the hard work will be proving prevention reduces incidents without creating new risk.