The productivity gurus on the internet these days argue you should be using AI agents to book travel, manage calendars, and send emails. Aslan emerged from stealth yesterday with a different take on what these little bots can do.
The DC-based startup raised $20.8M in pre-seed and seed funding, led by Khosla Ventures and XYZ Venture Capital, to build AI agents that take on fake personas—complete with a comprehensive, hyper-realistic digital footprint—and infiltrate encrypted criminal, terrorist, and adversary online networks on behalf of intelligence agencies and the US military.
That’s a new, very 2026 spin on the term “secret agents.”
Spybots: Aslan launched last year after CEO Chase Reid saw mass shootings and terrorist attacks planned in obscure online forums, and “if the information is out there on Discord, Telegram, or some other dark community, then we ought to find it before lives are lost.”
Traditional online intelligence gathering tradecraft focuses on “collecting as much information as possible, but in reality, you can’t collect from the outside of an encrypted group chat—you have to actually go inside,” he told Tectonic. “That’s the entire premise of Aslan. That’s exactly why this has to exist, because the adversary is going to continue operating in the dark, and if we’re not there with them, we’re not going to see, and we’re certainly not going to stop them.”
That’s a lot of work for human analysts, but Aslan’s betting that AI agents trained for this mission —and operating under the same legal guardrails as human secret agents—can help ease that burden. Here’s how it works:
- Aslan’s agents build up an online persona tailored for whatever (strictly) foreign groups they’re seeking to infiltrate, including by “architecting a fully holistic, culturally authentic background [to] build up this entire canon of backstopping so it can legitimately be invited into a community as a credible member,” Reid said.
- The agents, typically one or a small handful per intelligence or military analyst running them, operate in a disconnected physical device— “a physical Pelican case that includes all the equipment, all the computing for Aslan”—wherever the operator is based, instead of relaying the information gathered to the company. They’re model-agnostic and work across the American frontier AI labs.
- They also run their own agents to infiltrate foreign networks, including cybercrime groups and espionage operations, and present their findings to the US government agencies responsible for shutting them down.
Spy AI: While in stealth, Aslan has inked contracts with several intelligence community components, and they’re gaining traction with the US military and scoring some early wins.
- Their agents were key to identifying, infiltrating, and ultimately shutting down a Chinese government espionage program recruiting “defense-adjacent US professionals,” exposing a cyber-fraud marketplace, and mapping “a live cross-border smuggling network stretching from Mexican staging corridors into the United States,” according to the company.
- They also work with NORTHCOM to identify, track, and infiltrate cartel networks. One NORTHCOM analyst compared Aslan to “a giant ISR network” for the digital environment, Reid said. That info gathered can be used to build targeting packages.
“Any use case where a human operator needs to do the work manually, and there’s an online nexus, Aslan’s able to do it seamlessly and at scale,” he said. “It’s able to distill the most important mission-relevant findings and noteworthy developments, and then push that to the operator.”
- Reid estimates that Aslan has about “100 agents running out there in the world across different platforms, most of which are actually already burrowed in criminal groups, terrorist networks, and other types of really insidious forums online.”
Cash money: With nearly $21M in funding, Aslan’s looking to build out their team, run more agents and foreign infiltration operations themselves, and expand work with the US military and IC.
“The beauty of our Aslan agents is that as they spend more time in these ecosystems, not only are they learning the vernacular and the tactics so that every subsequent agent is that much better at infiltrating, but they’re also going to persist through the platform bans, through the reconstitutions, through the rotation of the names and identities of the people within those ecosystems,” Reid said. “They’re almost a permanent presence in places we otherwise would have zero visibility into.”
