40 +
Companies running on Lia
Tools that take routine work off your plate and give people time for what actually matters
Companies running on Lia
Engineers and product leads on the team
Dialogues handled by AI each month
We change plans in a day, ship in a week, rewrite the architecture in a quarter. That's how we keep pace with the market since 2017.
No release ships without a metric. We measure customer savings in money and FCR — not demo applause.
The bot handles routine. The agent handles complex cases. A human confirms critical actions — refunds, customer emails, data changes.
Lia takes routine work off teams — calls, chats, repetitive operations. The business gets back the hours its people used to spend on the same tasks every day.
Lia has automated communications and business processes for large enterprises since 2017. We build AI agents on neural networks and NLP tuned for the realities of enterprise work. 50+ engineers, product leads, and AI researchers — with backgrounds in banking, retail, and foodtech.
Stack: LLMs (Yandex GPT, GigaChat, open-source), RAG, on-prem. Systems hold peaks above 5,000 RPS at 99.98% uptime. Customer data runs through isolated contours, with an audit log for every agent call.
Lia closes first- and second-line support, sales, marketing, and back-office operations. On current customers — Samokat, Dodo, BetBoom, Whoosh, Localrent — pilots pay back from 2 months.
The company is founded. The whole year goes into building an MVP for banking support automation — the founders come from banking, and the first pilot is DeloBank. No paying customers yet.
First major banking customers — Rocketbank and VTB Light. We work an integration model: every project is built per customer, no no-code builder, no automatic dialogue classification.
Investment round from the VkusVill fund. Team refresh and a pivot to growth: sales scale up, the first non-banking customer ships — a pilot with Delivery Club.
We step away from banking and the tender model: integrations were too heavy and too slow. Pivot toward fast-cycle customers — retail, e-commerce, delivery services.
Catching up to the market and rebuilding the architecture: move to Kubernetes and managed databases, lift in stability and performance. Late in the year — first production RAG launches and a low-confidence classifier; Lia takes on first- and second-line support.
Doubled spend on marketing and sales, ran dozens of bets, hardened the platform for traffic peaks. In parallel — the speech-analytics module went into development and the foundation of the AI-agents platform took shape.
Speech-analytics module is in production — fully LLM-driven with a fast classifier, first customers live. New modules Magic View and Magic Query are out, the AI-agents platform is in beta with first design-partner teams.
Owns strategy and product vision.
Ex-CEO & Co-founder of Zebrainy (acquired by Yandex). 18+ years in IT — launching and scaling products.
Demian Greenberg
CEO & Founder
Owns product strategy and the metrics that prove it.
18 years in IT. Founded four companies (Lia among them) with a combined valuation of $20M. Built four fintech products from scratch, including Vysoky Bank and Modulbank.
Kirill Paraska
CPO & Founder
Owns engineering and platform architecture.
Built the engineering org at Vysoky Bank. 11+ years in IT products.
Nikita Levashov
CTO & Founder
Owns business development and large-account work.
Ex-bizdev at Yandex. 10+ years in sales, 6 in IT.
Andrey Nadvornyy
CBDO & Co-founder