DesignVerse targets U.S. firms with private cloud AI tools

Romanian-founded software firm DesignVerse has begun expanding into the United States, targeting security-conscious enterprises in aviation, defence and financial services.

The company builds its proprietary "enterprise context layer," which turns a client's engineering and business data into structured context for large language models (LLMs), enabling them to generate code consistent with existing architecture, technical standards and internal rules.

The European company has newly growing roots, founded roughly eighteen months ago by Chief Executive Andrei Manolache and Robert Drăguțoiu (Chief Technology Officer). The business initially focused on visual prototyping and A/B testing before shifting toward a full end-to-end offering as the market moved towards coding agents and assistants in an integrated context layer.

Manolache said that, as of now, around 90 per cent of its customers are based in Europe. The company currently has two customers in the U.S., which it describes as an emerging market for the business.

"We have seen at least in the last three months, with all of the talks from the White House banning Fable from Anthropic and then bringing it back, and then OpenAI essentially cutting access to their LLMs to Cursor because they got acquired by [SpaceX], that the market is starting to shift towards more providers who can offer this on-premise custom tailored solution to them," added Manolache.

The company's approach to new markets involves building a client base within a single industry, generally through partnerships with systems integrators, before establishing direct sales efforts and moving into adjacent sectors. Manolache said DesignVerse plans to apply the same approach in the U.S. that it has used in Europe, before eventually extending into healthcare, insurance and supply chain management.

"The reason why we're targeting the U.S. market right now is that we think that the market is finally starting to mature, and realising that AI is already very good at writing code, but for those large enterprises with very high reliability stakes, you cannot just use generalistic tools to solve your problem," Manolache added. "AI needs to understand how you do code writing from your architecture perspective, what you have built upon for the last 10, 15, 20 years, but also how this plays out with your existing business rules."

Manolache argues argues that regulated buyers currently have few options beyond routing requests through external model providers, potentially limiting their visibility into where proprietary data travels. In his framing, a zero-retention policy is a promise rather than an architecture; it is not the same as an enforced air gap.

DesignVerse's answer is a model deliberately built small enough to run on the client's own hardware, fully sealed so a CIO can account for every byte in and out. 

DesignVerse's client base is concentrated in high-reliability sectors including aviation, financial services and, increasingly, defence. Manolache said the company has worked with European air traffic management organisation EUROCONTROL since January this year, beginning with internal modernisation work of its Integrated Network Management program.

"In Europe, we're starting to do on-premise deployments with a bunch of our clients, which means that they have full control over their data - what goes inside those LLMs, what comes out of those LLMs," said Manolache. "The LLM that we have created is small enough that it can actually be deployed on the client's hardware in their building, completely sealed. They know what information goes in, what information goes out."