Concrete cases where we built AI infrastructure with measurable economic impact.
Online pharmacy (EU) · AI marketing · GDPR
AI marketing and reactivation for an online pharmacy
Value: retention and reactivation automation through customer segmentation, triggered campaigns, and personalised communications across CRM and the data warehouse.
How we apply AI: a closed AI perimeter with a fail-closed PII guard, an outbound allow-list, EU data residency, and provider-side zero retention. Campaigns pass an offline evaluation harness before release.
Impact: marketing scenarios receive automated regression checks before launch while personal data never leaves the perimeter in clear text.
Telecom · CRM business analysis · On-premises AI
AI business-analysis perimeter for an enterprise CRM rollout
Value: accelerate telecom CRM business analysis by reconstructing as-is processes, RACI, counterparty registers, and glossaries from email, chat, and document systems.
How we apply AI: an on-premises privacy perimeter with multimodal scan recognition, personal-data de-identification, reversible-token vaults, RAG, and adversarial verification of low-confidence facts.
Impact: business-analysis artefacts carry confidence scores and traceability; personal data and trade secrets remain inside the perimeter.
Video surveillance + access control · Multi-tenant SaaS
AI-native SaaS platform for video surveillance and access control
Value: cameras and access controllers from multiple vendors connect through unified adapters, with billing, analytics, and roles for many independent tenants.
How we apply AI: schema-per-tenant database isolation, role-based access, vendor-agnostic adapters, and an automated architect-plus-executor deployment workflow.
Impact: one codebase serves many tenants under strict data isolation, with releases executed through an orchestrated pipeline rather than manually.
Full case study: cloud for meaning, edge for frames →
Fintech · B2B platform · Greenfield development
BelVEB BusinessHub — a secure B2B transaction platform
Value: automate the full B2B transaction lifecycle with verification, security, and banking-instrument integrations.
How we apply AI: greenfield architecture, banking-system integrations, and a full lifecycle from design through launch within the bank perimeter.
Impact: a platform for Bank BelVEB supports verified sellers and buyers and secure transactions using banking instruments.
belvebhub.by
Science · Multi-agent AI · RAG
AI agent for anticancer-compound research
Value: accelerate scientific research through structure modelling, literature analysis, and automated review drafting across several research teams.
How we apply AI: a multi-agent RAG system on corporate data, semantic search, review generation, and ligand–target interaction modelling.
Impact: a high-accuracy multi-agent system for private laboratory data in organometallic anticancer research.
Healthcare · RAG · On-premises LLM
Clinical AI assistant based on RAG + LLM
Value: real-time clinician support using clinical guidance and a local patient-history database, with no data sent to external APIs.
How we apply AI: a Telegram bot, local embeddings, RAG search across internal documents, and image analysis of X-rays and reports.
Impact: the design validates the importance of keeping patient data inside the company for privacy and security.
seraph.ai-automation.llc
Healthcare · Scale-up
Talon.by — digital medical appointments
Value: national-scale healthcare digitisation through a unified electronic-queue platform, paid-service enablement, and government-system integrations.
Impact: the project has been deployed at more than 1,000 healthcare facilities across the Republic of Belarus.
International startup · AI optimisation
Kuda.party — a leisure-discovery platform
Value: optimise a startup’s development across iOS, Android, and RuStore with AI tools that accelerate the team and lower cost.
Impact: development costs are three times lower than comparable market offerings, and the application is available in three stores.
Oil trading · Document workflows · NDA
AI automation of primary-document processing
Value: automate incoming document workflows: collect offers, analyse them semantically, match them against a database, generate natural-language briefs, and export to CRM.
How we apply AI: an AI pipeline with knowledge-base matching, natural-language search, and a chat-based operational interface. Data remains inside the perimeter.
Impact: counterparty-offer processing is removed from operators, and a pilot on real client data is possible.