IdaProject: AI could cut developer layout-rendering costs by 50-60% at scale

September 2, 2026 · cossaru

Pavel Gavrichenko, design and product director at IdaProject, lays out the arithmetic behind manual real estate copywriting: a twelve-person team at one developer needed three to four months to write descriptions for 50 residential projects, not counting individual apartment cards. Writing a single floor-plan description takes about two hours, plus two to three hours for internal approvals, for up to six hours per layout, and a run of hundreds of layouts can stretch the cycle to five or six months. IdaProject built a system in which AI reads an architectural file, analyzes walls, room layout, door openings and bathrooms, groups apartments by shared features such as balconies or window orientation, a task that now takes about an hour, and drafts a listing description in 20 seconds; a product specialist then checks the text for accuracy and brand tone. Rendering and describing one floor plan the classical way costs 1,500 to 4,000 rubles, and for one top-25 ERZ developer redrawing 20,000 floor plans, IdaProject projects AI would cut what it calls an astronomical classical-approach cost by 50 to 60 percent.

Search algorithms on Cian, Domclick and Avito favor unique listing content, so developers who give every unit its own AI-generated description can see card rankings and view counts improve on these marketplaces, a factor agencies buying real estate media there should track. IdaProject also argues AI removes the idle-cost problem of paying copywriters or an outside agency through a developer's low season, since the AI system needs no salary when there is no new inventory to describe. The HubSpot State of Marketing study cited in the piece found 85 percent of marketers say generative AI is transforming content production while 82 percent report a rise in output capacity, a pattern that matches what IdaProject describes for property listings and one that agencies serving real estate clients can use to benchmark their own production costs against AI-assisted alternatives.