Specific deployments with before/after numbers, stack and architecture details. Grouped by 4 service tracks. More under NDA — I'll show them on a call.
Nothing matched. Try another word — or just message me, the task is probably doable anyway.
Voice agents, 24/7 booking, reputation management. Who it's for: the small-business owner losing enquiries after 7 PM.
The methodology is ready, but I haven't yet run client projects in these three niches — salons, dental clinics, auto shops. The technical stack is thought out and tested in my adjacent deployments (Vapi, GigaChat, WAPI, YClients and 1C integrations). For the first clients in these niches — preferential terms: −30% off the price, extended 90-day warranty, and the right to describe the project without naming the company. The numbers in the cards are targets based on typical 2026 benchmarks for Russian small and mid-sized businesses.
Picks up calls 24/7, books a slot in YClients, sends SMS confirmation. 14 scenarios; when the model isn't confident in its answer, it hands the call to an admin.
Voice + chat with prioritization: acute pain / scheduled / prophylactic. FZ-152: data in RU, GigaChat without identifiers.
Cloud-telephony webhook → look up the customer and their order in Shopify → SMS with the actual ship date. Every miss becomes a CRM task.
Own RAG engine over a product fact corpus: SSE streaming, voice input, anti-abuse and grounding tests. Catalogue answers are locked down by deterministic guards, and the lead reaches the sales manager with the whole conversation attached. The paid licence was switched off.
Lead funnel, marketing automation, lead generation, attribution. Who it's for: heads of sales, marketing or growth.
One n8n workflow across 8 clients: pulls the numbers from every ad platform, AI commentary, PDF/Notion → auto-sent Mon 09:00.
Orders, tracking numbers, statuses and cancellations land by themselves every 30 minutes. An Advanced Sheets Service receiver inserts rows mid-grid and carries formulas across; reconciliation and backups run on timers.
1C and enterprise-system integrations, document processing, internal knowledge-base search, local models, networking, AI security audit, process optimization. Who it's for: CTOs, CIOs and heads of IT.
XML protocol V3 + V4 → 1CClientBankExchange format, Windows-1251.
2-week audit: shadow sessions across 6 roles, time-tracking of top-20 operations. Found: auto-sync of stock across WB/Ozon/1C + AI first line of support + auto-generation of UPD.
2.5-week red-team per OWASP LLM Top 10. Findings: 8 critical + 11 high + 14 medium. Grounds for an FZ-152 fine removed. Passed pre-IPO DD with no findings.
aiogram 3 + Node.js middleware via 1C HTTP services. Access to 1C UNF from a site without a laptop: order statuses, budget left, material confirmation via photo.
BM25 → SQLite FTS5 with a custom ranker, voice transcription, OCR, access tiers and an MCP server over the knowledge base. The index went from a gigabyte in RAM to tens of megabytes.
Websites with telephony and a map listing, mobile apps (Swift / SwiftUI), PWAs with iOS Add to Home Screen, playable HTML5 for performance campaigns, full-stack web apps (Next.js + FastAPI). Track record: 20+ iOS apps in Swift, engineering experience since 2017, production SaaS platforms with multi-user RBAC.
Native iOS 17+: schedule + push (24h/1h/10min), Speech Framework for on-device pronunciation, StoreKit 2 + YooKassa, Firebase backend. App Store + RuStore.
PWA with iOS Add to Home Screen, offline via Service Worker, WebPush iOS 16.4+, YClients embed + YooKassa with no 30% Apple commission. ~890k ₽/year saved.
A custom mail indexer over IMAP, a customer card with order history from both stores, LLM-drafted replies. Opening a card got three times faster.
Not every project can be published. On a free 15-minute call I'll walk you through the architecture of 2–3 closed deployments for your industry — without client names, but with numbers and technical details.
Book a callI'll tell you what to deploy in your business first, what the payback looks like, and whether you need AI for the task at all (sometimes you don't).
Or just send your question — I reply within 2 hours