Eighty-Five Million Dollars for a Company With No App to Show You

What if the most valuable AI companies of the next decade don’t ship a chatbot, an app, or anything you can click at all?

That question is worth sitting with, because a startup called Hang Ten Systems just added $53 million to its seed round, bringing its total funding to $85 million. It was founded by Vishal Sikka, the former CEO of Infosys. The new money was led by Temasek’s Xora, with Mayfield and Aramco Ventures joining in. And the description of what the company does is three words long: enterprise AI services.

If you’re used to AI news arriving in the form of a demo video, that’s an unsatisfying answer. So let’s unpack why serious investors are writing eight-figure checks for something so hard to screenshot.

Services versus products, in plain terms

Most AI companies you’ve heard of sell a product. You sign up, you get access to a thing, the thing does something. The company builds one version and sells it to millions of people.

A services company works differently. It goes inside one organization at a time and does the work of making technology actually function there. Less “here’s our app,” more “we’ll come figure out your mess.”

That distinction matters enormously for AI agents specifically. An agent isn’t a tool you point at a task once. It’s software that takes actions on your behalf across systems, over time, with some degree of independence. For that to work at a large company, the agent needs to reach into the customer database, the ticketing system, the inventory records, the approval workflows, the twenty-year-old application nobody wants to touch. It needs permissions. It needs guardrails. It needs someone to decide what happens when it gets something wrong.

None of that arrives in a box. It gets built, per company, by people who understand both the AI and the company. That’s the work a services business does.

Why a former Infosys CEO is an interesting bet

Infosys is one of the largest technology services firms in the world. Its business, broadly, has been going into big organizations and doing their technology work: building systems, running systems, staffing the humans who keep it all upright.

Someone who ran a company like that has a specific kind of knowledge that’s genuinely hard to acquire. Not “how do I train a model” but “how does a bank with forty thousand employees and decades of accumulated software actually adopt something new without breaking payroll.” Those are different skills, and the second one is arguably the bottleneck right now.

Plenty of companies have access to capable AI models. Fewer have any idea how to wire those models into the daily operation of a business that already works a certain way. If you believe the shortage is deployment rather than capability, funding a services company founded by a services executive starts looking less strange.

What the funding pattern tells us

The mechanics here are unusual enough to notice. This is an expansion of a seed round, not a Series A. Seed funding is normally the earliest and smallest money a startup raises, meant to get an idea off the ground. Eighty-five million dollars at seed stage means investors committed before there was much of a track record to evaluate.

That happens when investors are betting on the founder and the timing rather than on results. It also happens when the thing being built requires real money up front. A services business that needs experienced people in the room with enterprise customers is expensive from day one. You can’t run it lean out of a spare bedroom.

The investor mix is telling too. Temasek’s Xora led. Mayfield and Aramco Ventures participated. That’s a blend of Silicon Valley venture capital and money connected to large industrial and institutional interests, the sort of organizations that would themselves be customers for enterprise AI work.

What this means if you’re not in the room

For most readers, the practical takeaway isn’t about Hang Ten Systems at all. It’s about where the AI agent story is heading.

• The interesting money is moving toward implementation, not just capability. Getting agents to work inside real organizations is its own hard problem.
• Agents are being built to operate inside existing business systems rather than replace them. That’s slower and less flashy than a new app, and probably more consequential.
• Companies that look boring from the outside can be doing the load-bearing work. Not everything important has a user interface.

The agents that end up affecting your job most may never have a logo you recognize. They’ll be running quietly inside the systems your employer already uses, installed by people whose entire business is knowing how to do that carefully. Eighty-five million dollars says at least a few investors are betting on exactly that.

🕒 Published: September 17, 2026