The Stanford engineer who joined Virtue AI to sell it - and three months later found himself running the whole thing.
In April 2026, Sohaib Shaikh walked into Virtue AI with a familiar job title - Head of Sales. It is the sort of role that usually comes with a long runway, a quota, and the quiet understanding that the corner office belongs to someone else. By June, the corner office belonged to him. Virtue AI named Shaikh its Chief Executive Officer, and the story of the year in AI security gained an unexpected protagonist: not a researcher, not a founder, but a closer.
There is a certain comedy in this. Virtue AI was founded by four names that make academics go weak at the knees - Bo Li, Dawn Song, Carlos Guestrin and Sanmi Koyejo, a bench of AI safety scholars whose citation counts read like phone numbers. They built the science of keeping artificial intelligence from misbehaving. And then, when it came time to decide who would actually run the company, they handed the keys to the person best at explaining why a Fortune 500 chief information security officer should sign the check. Brilliant research that nobody buys, after all, is just an expensive hobby.
Shaikh's arrival at the top coincided with a number that tends to end arguments: 600 percent year-over-year growth. When a company grows that fast, it does not need a caretaker. It needs someone who can keep the wheels attached while the car accelerates, and who understands - from the buyer's side of the table - exactly why enterprises are reaching for a product like this in the first place.
His answer to that question is disarmingly plain. "Enterprises are deploying autonomous agents faster than they can secure them," Shaikh has said. "Our platform provides the trusted foundation they need to keep up." It is not a line built for a TED stage. It is a line built for a procurement meeting, which is precisely the point.
Enterprises are deploying autonomous agents faster than they can secure them. Our platform provides the trusted foundation they need to keep up.- Sohaib Shaikh, CEO, Virtue AI
Before Virtue AI, Shaikh spent more than eight years at C3 AI, the enterprise AI company where he rose to Group Vice President of Sales. Eight years in one place is a long time in Silicon Valley, where tenure is often measured in vesting cliffs. What it bought him was a rare and unglamorous education: how large, cautious, heavily-regulated organizations actually decide to adopt artificial intelligence. Not the demo. The signature.
His resume before that reads like a series of unrelated postcards - stops connected to Warner Music Group, Oracle, and Stanford University, where he earned a Master's in Electrical Engineering between 2016 and 2017. Read casually, it looks scattered. Read carefully, there is a through-line: at every stop, the assignment was to take something technically dense and make a serious buyer say yes. Music catalogs, database software, enterprise AI - the products changed, the skill did not.
That skill is easy to underrate. The technology press lionizes the people who write the papers and file the patents. Far less ink is spent on the people who translate that work into revenue, and yet without them the papers stay on the shelf. Shaikh belongs to the second tribe, and Virtue AI's decision to make him CEO is a small vote for the proposition that in enterprise AI, distribution can matter as much as invention.
Earns a Master's in Electrical Engineering from Stanford University.
Begins an eight-plus-year run at C3 AI, rising to Group Vice President of Sales.
Joins Virtue AI as Head of Sales, leaving enterprise software to bet on AI security.
Named CEO as founders step back and the company posts 600% growth.
Bars are illustrative, scaled for comparison.
Virtue AI calls itself the trust layer for AI and agentic systems - the software that sits between an enterprise and the increasingly autonomous models it wants to deploy, checking that they behave. The pitch is simple: security should be the thing that lets companies move faster, not the thing that slows them down.
Founded by AI safety researchers Bo Li, Dawn Song, Carlos Guestrin and Sanmi Koyejo, the San Francisco company raised $30 million in combined seed and Series A funding, led by Lightspeed Venture Partners and Walden Catalyst, with participation from Prosperity7 and others. Its architecture bundles automated red-teaming, real-time multilingual and multimodal guardrails, and governance for enterprise agents into a single stack.
Real-time guardrails that screen chatbot and agent traffic before it reaches - or leaves - your users.
Continuous, automated red-teaming that probes production AI for weaknesses the way an attacker would.
Security, governance and a testing ground purpose-built for autonomous agentic systems.
Endpoint-level discovery and monitoring to find the AI agents already running inside the building.
The leadership transition that installed Shaikh also saw founders Bo Li and Sanmi Koyejo step away from the company, while Daniel Le - a former CFO of dbt Labs and finance chief from Zoom - stayed on as Chief Financial and Operating Officer. Read together, the moves sketch a deliberate shift: from a research-led startup proving what is possible to a commercially-led one proving what will sell.
It is a bet with real stakes. AI safety is one of the hardest technical problems in the industry, and there is a fair question about whether a sales leader is the right person to steer it. The counterargument is that the science already has custodians in the founding team and advisory bench - the missing ingredient was someone who could carry it into thousands of enterprise deployments. That is the job Shaikh has spent his entire career rehearsing.
What he inherits is momentum and a thesis he clearly believes: that the gap between how fast companies deploy AI agents and how fast they secure them is widening, and that whoever closes it owns a large and durable market. It is not the most glamorous story in artificial intelligence. It may turn out to be one of the most important.