Lazarev wants your AI to survive Monday morning

The revealing detail in Lazarev’s Fieldstream case is that people kept leaving. Fieldstream could analyze marketing performance and model future spending. Yet, according to the agency’s account, users exported data to make decisions somewhere else. The software had produced an answer. The customer still had a job to finish.

That is an awkward problem for an AI company. A demonstration can end at the moment a chart appears. A working day continues: someone has to compare alternatives, impose a budget limit and explain the choice to a colleague. The distance between those two moments is where Lazarev has planted its business.

The Ukrainian-founded agency designs websites, apps and digital products, with a particular emphasis on AI and business software. Its portfolio is full of visual polish. Look underneath, though, and a more interesting ambition emerges: keep the user’s work from falling through the cracks between screens.

01 The customer’s other window

For Fieldstream, Lazarev describes reorganizing an analytics product around the decision the marketer needed to make. Filters became more consistent. Overloaded views were separated. Planning acquired a visible sequence: establish the plan, add constraints, choose among the options.

The distinction matters because a dashboard is a place to look, while a planning tool is a place to commit. A marketer needs to know what changing a control will affect. A forecast needs to sit close enough to its assumptions that someone can challenge it. These are small interface decisions with consequences for whether a product can be used without an interpreter.

The useful question for a product team is observational: what is open in the next browser tab? If the real decision happens there, the product may be outsourcing its most valuable moment. Fieldstream gives that question a concrete setting, even for readers who will never buy marketing analytics software.

02 A business education, paid by the website

Kirill Lazarev founded the agency in 2015. In his own telling, website design offered a way to learn entrepreneurship while earning money from entrepreneurs. He later faced a digital state-infrastructure assignment large enough to require help from experienced designers. The business began to become a team.

“It was a perfect combo to learn from entrepreneurs first-hand and make money.”

Kirill Lazarev / on his early website work

There is an appealing practicality to that origin. The client supplies a business problem; the designer gets an education in how that business works. Today the agency sells a much broader service, but the same arrangement is recognizable in projects that demand an understanding of research, marketing or customer acquisition before a screen can be useful.

Its Ukrainian history also explains a project that does little to resemble enterprise AI. After Russia’s full-scale invasion, Ukrainian Power presented creative agencies to international buyers. Communication Arts describes a directory of more than 30 agencies employing over 770 people between them, designed in Figma and built in Webflow. Lazarev was helping other firms get noticed, including firms a buyer might choose instead of Lazarev.

In 2022, its own website won the Webby People’s Voice award for Professional Services & Self-Promotion. At the ceremony, the founder used his five-word speech for Ukraine. The agency’s public story about design and business sits inside a larger history of people trying to keep working.

03 A conversation needs somewhere to go

Accern’s Rhea offers a different test. Lazarev says its relationship with Accern started in 2022 and expanded in 2024 to a research product for financial analysts, venture investors and ESG specialists. These users need to inspect material, compare it and make something another person can read.

A chat transcript is an uncomfortable container for all of that. Rhea’s design pairs prompts with graphical controls, references and charts. A split-screen reporting mode lets the researcher assemble an output while continuing the investigation. Dataset selection and file management supply the context around the conversation.

The case’s rationale is more persuasive than the familiar promise to make AI intuitive. Researchers work with tables and graphics; therefore, the interface must accommodate them. That is a design choice tied to a particular job, rather than a rule that every product should look like a messenger.

Lazarev’s separate writing on AI design adds a useful principle: outputs should carry references and controls, and should be reusable in the next stage of work. A fluent answer is only one component. The researcher also needs ways to inspect it, adjust it and carry it forward.

The portfolio attaches a $40 million-plus client fundraising figure to the years-long Accern relationship. Read that as the agency’s account of a client milestone. It does not isolate how much the interface contributed. The split-screen report workflow is the part another team can actually examine and borrow.

04 From financial reports to the camera roll

The same agency also worked on Elva, a mobile AI video editor. Its premise is easy to recognize: phones accumulate footage faster than their owners edit it. Elva takes a spoken request and turns camera-roll material into a video. Lazarev’s case covers the brand, interface, onboarding and monetization as one assignment.

Here the design problem shifts. A financial researcher wants room to investigate; a casual video maker wants less editing labor. Elva’s visual persona and onboarding help introduce that bargain. The Webby organization lists Elva as the 2026 People’s Voice winner in AI Visual Design. That is recognition of the experience’s presentation, a different measure from how many people keep using it.

This range helps locate Lazarev in the market. It is a services business that can connect product behavior with how the product is introduced and sold. Customers commission work; they do not buy a Lazarev subscription to Rhea or Fieldstream. The useful unit of comparison is the agency engagement.

05 The price of a substantial assignment

On Clutch, Lazarev currently lists a minimum project size of $100,000 and hourly rates of $100-$149. The same profile contains historical reviews of smaller projects, with $10,000-$49,999 the most common reviewed bracket. Those numbers describe different things: today’s advertised entry point and an accumulated record of earlier assignments.

Time belongs in the budget too. Lazarev describes phased engagements, with an initial focused experience taking roughly eight to twelve weeks and further phases expanding the work. It argues for doing discovery properly before narrowing the first design slice. That proposition suits a company with a meaningful workflow problem and enough organizational attention to address it.

Alternatives depend on the assignment. Clay combines branding and digital product work. Eleken offers subscription-based SaaS design. A company with daily, deeply internal needs may prefer its own designer; a narrow website job may fit an independent specialist. Lazarev’s AI cases give buyers a specific body of work to test against their problem.

06 Borrow the detour

The cheapest useful lesson here requires no agency contract. Watch a customer finish one valuable task. Include the export, the second application, the colleague who explains the chart and the document eventually sent. Mark the point at which the product stops helping.

Then design a smaller connection across that gap. Preserve a filter. Put an explanation beside a chart. Let a person compare plans before committing. Give a generated answer somewhere useful to go. These suggestions follow from the cases; they are experiments to test with users, rather than guaranteed results.

There are limits. A polished interface cannot repair unreliable underlying data or create demand for an unwanted product. The agency itself points to early product-market fit and measurable data as favorable conditions. A redesign needs something real to improve and a way to tell whether it did.

Lazarev’s most interesting claim on a buyer’s attention is therefore quite ordinary. It has spent time thinking about what happens after the impressive part. Monday arrives. Somebody opens the software. There is still a report to send.