Yury Koretskiy — AI Consulting & Automation
I build AI systems
that actually run.
Practical AI for businesses that want working software, not slideware. Strategy, custom builds, and the plumbing in between — connecting your tools, data, and decisions.
What I do
Most businesses know they need AI. Few know where to start.AI Consultancy
Which tools, where to start, what to skip. Strategic guidance grounded in systems I've actually built — not vendor brochures.
Custom AI Solutions
Chatbots, agents, and data pipelines — designed, built, deployed, and maintained on your stack or mine.
AI for Marketing
Content engines, lead generation, and campaign automation that run while you sleep.
Workflow Automation
Your tools finally talking to each other. n8n, APIs, and glue code that holds under load.
Personal AI Assistant
Scheduling, research, and email handled by an assistant wired into your actual workflow.
AI Audit
A hard look at your current AI usage: what to keep, what to fix, what to kill.
Shipped systems
A selection from the ledger. Every row is real and running.| No. | System | Function | Stack | Status |
|---|---|---|---|---|
| 001 | Push Notification Service | Self-hosted alerts from any system to any device. | Next.js · Drizzle · SQLite | live |
| 002 | AI Chat Assistant | Customer conversations answered by AI, escalated to humans when it matters. | Chatwoot · n8n · OpenAI | live |
| 003 | Cash Tracker | Personal finance tracking with live balance updates. | Node.js · Docker | live |
| 004 | Personal Dashboard | One screen watching every running system. | nginx · Docker | live |
| 005 | HyperFrames | Programmatic video engine — HTML in, finished MP4 with voiceover out. | Node · Whisper · TTS | engine |
| 006 | Trendwatching Pipelines | Multi-platform trend scrapers with scoring and HTML dashboards. | Python · Apify | pipeline |
| 007 | Semantic Bookmark Search | Vector search over a personal knowledge base of thousands of links. | Supabase · pgvector | system |
| 008 | Messenger Postback Bot | Conversational lead capture wired straight into the funnel. | Python · Flask · Docker | live |
+ 60 more entries in the ledger — automation agents, internal tools, research pipelines. Ask me about any of them.
How I work
Practical over hype
Start with the problem, not the model. If a spreadsheet solves it, you'll hear that first — and it builds the trust that matters when AI is the answer.
AI reasons, code executes
Probabilistic AI for judgment, deterministic code for execution. That separation is why my systems stay reliable after the demo ends.
Ship, then sharpen
A running v1 this week beats an architecture diagram next quarter. Everything in the ledger above started as a one-week build.
"The gap isn't between companies that use AI and those that don't. It's between companies with running systems and companies with plans."
04 — Contact
Ready to start?
Tell me the problem. I'll tell you honestly whether AI solves it — and if it does, I'll build the system that proves it.