Two AI startups raise a combined $350 million for personal assistance and AI security

Instinct, a San Francisco-based startup behind a personal AI agent, raised a $250 million Series B round on 26 August 2026. The round valued the company at $2.5 billion, up from $50 million just four months earlier. The round was led by Index Ventures and Benchmark, with Conviction, Greenoaks and Kleiner Perkins also participating. The company has now received a total of $350 million in funding.

On 2 September 2026, Austin, Texas-based cybersecurity company HiddenLayer announced its own $100 million Series B. That round, led by Delta-v Capital with participation from Morgan Stanley, Microsoft's M12 and Booz Allen Hamilton, among others, brought total funding to just over $155 million. Together, the two rounds illustrate where investors are currently putting their money: AI agents that carry out tasks autonomously, and the infrastructure to keep those agents secure.

Instinct wants to take over tasks, not just provide answers

Instinct operates formally under the name Spear Street Technology and was founded by Noah Shinn, a 23-year-old former researcher at enterprise AI startup Sierra. The product is an AI agent that communicates via text and phone calls and connects to the user's apps and accounts, including email, calendars, messaging services, phone and computer.

The agent can handle a wide range of tasks: planning travel, ordering groceries, buying concert tickets, cancelling subscriptions, responding to emails, booking rides, comparing insurance policies and even helping to rent an apartment or organise a wedding. This goes beyond a conversational interface: the agent completes transactions in external systems.

For now, the product is available exclusively through an invite-only beta programme. The deep access to personal data also raises privacy questions. Instinct gives users the option to opt out of having their data used in model training, but the broader discussion about what it means to grant an AI agent access to sensitive accounts remains open.

HiddenLayer secures AI models and agents against attacks

Where Instinct focuses on what AI agents can do, HiddenLayer focuses on what goes wrong when they are not properly secured. The Austin, Texas company builds tools for vulnerability detection in AI models, runtime protection, attack simulations and AI supply chain security.

The threats HiddenLayer targets are specific to AI environments: prompt injection, agent manipulation and the exploitation of malicious tools introduced through the supply chain. These are attack vectors that rarely appear in traditional software security, but are structurally relevant in the context of AI agents.

Commercial traction is significant. Annual recurring revenue (ARR) increased tenfold over the past year and now stands in the tens of millions of dollars, with more than 90 percent coming from new customers. More than 50 new customers joined from sectors including banking, insurance, pharmaceuticals and aviation. HiddenLayer also works for US government agencies: the company was selected in January 2026 for the US Missile Defense Agency's SHIELD contract and for the Department of Energy's Prometheus initiative.

With the new capital, the company intends to expand its sales and distribution capabilities, hire more engineers and establish a presence in Europe and the broader EMEA region.

Broader context: AI agents attract the lion's share of investment

The two rounds fit a clear pattern. In the final week of August 2026, 80 percent of the $1.59 billion in announced venture investments went to AI companies. Nearly a quarter of all new unicorns in 2026 (49 out of 191, or 25.7 percent) are active in AI.

Gartner estimates the global AI security market at nearly $4.8 billion in 2027, an increase of 68.7 percent compared with 2026. By 2028, that market is projected to reach nearly $7.7 billion. The same research firm predicts that more than half of successful cyberattacks on AI agents in 2029 will exploit weak access controls and prompt injection, precisely the attack vectors HiddenLayer targets.

For those building with AI or investing in it, the two rounds together point in a clear direction. Agents that autonomously perform actions in external systems rank high on investors' priority lists. At the same time, demand is growing for security infrastructure designed specifically for those agents, a market that barely existed until recently.