Vlad Medvedovsky wants AI to earn its keep

Software has acquired a remarkable talent for appearing finished. A screen lights up, a demonstration behaves, and a roomful of people briefly enjoys the agreeable sensation that the difficult part is over. Vlad Medvedovsky makes his living in the part that follows. Someone has to connect the systems, understand the customer’s business, and discover whether the elegant answer survives an ordinary working day.

As founder and CEO of Proxet, he works in the territory between technological possibility and operational responsibility. Data platforms, payments, cloud infrastructure, and AI all pass through that territory. The vocabulary changes with the conference season. The obligation to make something useful has a longer shelf life.

His latest proposition brings that obligation into AI-assisted software development. If machines can produce more code, the people directing them need a clearer account of what the code should accomplish. A very quick journey remains an inconvenience when the destination is wrong.

An engineer in someone else’s story

Medvedovsky’s background includes a bachelor’s degree in computer science from the University of Illinois Urbana-Champaign. His career has since crossed several of technology’s neighboring occupations: engineer, founder, investor, adviser, and the executive responsible when a client expects the work to arrive.

One revealing appearance comes in someone else’s account of starting a company. In November 2014, Dispatch co-founder Yaakov James Zar published an essay about bringing the business into public view. He described wanting to help entrepreneurs build their own products while relying on existing tools for the surrounding machinery.

Among his acknowledgments was Medvedovsky at Foundation Labs. Zar thanked him alongside the other co-founders and the team for helping make the product and launch happen. Public company profiles identify Medvedovsky as a Dispatch co-founder. The detail places him close to a familiar entrepreneurial problem: turning an idea into something another business can actually build upon.

There is a pleasing continuity here. Dispatch’s early argument concerned freeing customers to concentrate on their own businesses. Proxet’s work also enters the story at the point where another organization needs engineering help. The product differs; the customer’s wish to get on with things is recognizable.

His public investment record adds another connection to the startup world. Crunchbase identifies him as an investor in Catio and an adviser at Metadata. Those roles sit alongside his operating work at Proxet. They give the career several points of contact with young technology businesses: helping build one, helping deliver for others, and participating as an investor or adviser. The common setting is a company trying to turn technical work into a business.

The obligations that come with growth

At Proxet, that practical work has become an international organization. The company describes a team spread across North America, South America, and Europe, working in data, cloud services, product engineering, and managed teams. For a founder, that creates a second engineering assignment: arranging the people who deliver the engineering.

Proxet reported adding nearly 100 employees during 2024. In his year-end remarks, Medvedovsky described the continuing pressure of serving clients. It was a useful choice of emphasis. Hiring is an event a company can announce. Keeping promises is a condition it has to maintain.

“In the services business, we are constantly under pressure to provide outstanding service to our clients.”Vlad Medvedovsky, December 2024

The next spring brought appointments aimed at that responsibility. Dmitriy Bibikov joined as vice president of global delivery, leading the engineering department from Lisbon. Jason Price became director of account management, based in American Fork, Utah. Each brought about two decades of relevant experience.

Medvedovsky’s explanation centered on getting delivery and account management to work closely together. That pairing deserves attention. One function knows what the client expects; the other must organize the work that fulfills it. A promise can travel between departments with surprising changes of costume. Keeping its meaning intact is a substantial part of running a services company.

A seat at several tables

The public record of his work includes plenty of movement between industries. In March 2025, Medvedovsky attended Fintech Meetup with Ben Adair, Proxet’s fintech solutions lead. Their company’s account emphasized payments embedded in everyday software, the business opportunities around them, and the data needed to make them useful.

Those concerns are quite concrete. A software business adding payments also acquires transaction information, questions about customer behavior, and opportunities to improve reporting. Connecting a payment service begins a longer discussion about how the business will use what it learns.

By May, Medvedovsky was among the Proxet team at the inaugural EDGE Summit, where real estate and technology met. Colleagues included Dmitry Orlovsky, Gabriel Safar, and Igor Moochnick. Safar took part in sessions addressing AI, data, and partnerships. The group’s different specialties give a sense of the company around its founder: the conversations extend beyond a single executive’s technical vocabulary.

At the table in the event photograph, Medvedovsky sits opposite another attendee, with coffee cups and colleagues nearby. It is an unshowy image for a technology story, which is part of its appeal. Understanding a business still involves allowing another person to finish a sentence.

Powerful platforms still need plumbing

Proxet’s platform relationships give these conversations somewhere to go. In May 2024, it announced a partnership with Palantir. The work described included implementing Foundry and AIP, integrating enterprise information, establishing governance, building data models, and helping clients operate the resulting systems.

Medvedovsky framed the relationship around giving businesses of different sizes access to stronger control of their data. Palantir’s Ted Mabrey, then quoted as global head of commercial, welcomed the additional choice of development services for customers. The arrangement put Proxet in the implementation business around a platform with considerable capabilities of its own.

The distinction between owning a platform and using it well creates room for his company. An organization can acquire software while still needing help with the information, processes, and decisions surrounding it. The purchase order is usually shorter than the integration plan.

In September 2025, Proxet announced an Adyen partnership focused on payments. Its stated aims included simpler integrations, better use of payment information, and support for new products. Adyen’s own partner directory lists Proxet as a service partner available globally, with work extending from platform integration to data engineering and user adoption.

These relationships describe a particular position for Medvedovsky’s business. Customers bring their existing operations; technology partners bring platforms; Proxet supplies engineering and implementation. Much of the value has to be created in the connections, where the organizational diagram rarely looks as tidy as the sales presentation.

Before anyone presses generate

His thinking about data predates the current rush toward AI agents. In a 2021 discussion of data engineering, Medvedovsky described industries digitizing and the resulting changes in how companies acquire, clean, and process information. Businesses, he argued, would need internal capacity or dependable partners to build their data infrastructure.

That argument has an obvious relevance to AI work. A system asked to reason about a business depends on information that represents the business properly. Faster processing does little to settle a disagreement about what a number means. The spreadsheet has always had a talent for surviving its supposed replacements.

Proxet’s September 2026 launch of its Intent-Driven Lifecycle offering, shortened to IDLC, extends the concern from data to the delivery process itself. Medvedovsky’s accompanying statement puts the obstacle in human clarity and validation as code becomes easier to generate. The offering aims to organize AI-assisted work around common context, clearer problem definition, checks on results, and teams trained to direct the process.

Its commercial entry point is a two-hour executive workshop bringing together business, product, engineering, and quality leaders. A limited delivery stream follows, intended to establish the economics before wider adoption. The sequence gives the proposition a practical test: can the organization agree on an outcome and then demonstrate that the work meets it?

The framework also gives durable project knowledge a place in the process. Requirements, design decisions, and instructions are meant to remain available across AI sessions, with humans directing the work. Its published improvement figures are targets. The customer’s own delivery stream is where the economics must be established.

The same practical emphasis appeared in Proxet’s plans for Snowflake Summit in June 2026, where Medvedovsky was its listed attendee. The proposed conversations covered reliable analytics, production AI workflows, infrastructure costs, and engineers working alongside client teams. These are different ways of asking whether a platform is serving the organization that pays for it. For Medvedovsky, appearing at the summit placed the founder directly in those discussions, alongside the company’s broader effort to help clients turn their data infrastructure into working systems.

The next conversation

The company’s October 2026 listing in the Claude Partner Network shows where that approach is being offered: commercial real estate, investment and wealth management, and other software-intensive businesses. Its description connects data preparation with applications such as document processing and reporting. These are settings where a useful answer needs to fit the information and tools already in use.

Medvedovsky’s own public activity keeps returning to the people doing that work. His announced plans for FuturOps in Dallas on October 26–27 invited conversations about projects built with Palantir. The question behind the invitation was straightforward: what are you building, and where is it getting stuck?

It makes a fitting current chapter for someone whose career includes helping another founder get a product launched. The technology has become more capable. The founder’s assignment continues to involve understanding someone else’s ambition well enough to help make it work. AI can now join that conversation. It still needs somebody to explain the job.