I build AI-first companies — and set up the same AI system for other business owners.
I run two companies, Andriichuk OÜ in Estonia and Gazolin Production in Romania, where AI handles research, code, support drafts and admin, and I make the decisions. Through AI in Order I set up that kind of AI operating system for owners who run several clients. Before that I spent ten years building data and AI systems for PIN-UP / RedCore, Deloitte, PepsiCo and Health Union.
Companies, products and systems built AI-first.
AI-first companies
I build companies where AI is part of the operating model from day one. Research, code, payments setup, support and outreach run through one AI system with a shared memory. The team stays small and every decision stays with a person.
SaaS and micro-SaaS
Narrow products that do one job well, from QuickBooks file conversion to Chrome extensions and vertical B2B platforms. Each idea starts as a landing page with a price; I build the product once real people ask for it.
AI operating systems for owners
Through AI in Order I set up a personal AI operating system for owners and agencies: Claude and ChatGPT, a memory in plain files, Telegram and connectors to the tools they already pay for. Mainstream tools only, no fragile custom code.
aiinorder.com →Two companies, one way of working.
Each company has its own site with products, legal details and contacts.
Andriichuk OÜ
An independent micro-SaaS studio. Small focused products for bookkeepers, media buyers, meetings and everyday browsing, plus the AI in Order service.
andriichuk.com →Gazolin Production
A Romanian company running five focused businesses: AI voice agents and chatbots, AI player support for iGaming operators, AI automation for regulated finance, content and performance marketing, and a Land Rover specialist service.
gazolin-production.com →I set up for clients the system I run my own companies on.
AI in Order is not a theory I sell. It is how my companies work every day, packaged so that an owner without an engineering team can have it too.
- One AI system for the whole business. My companies run on Claude with a memory in plain files, connectors to mail, calendar and the browser, and repeatable skills for routine work: launches, payments setup, legal pages, outreach.
- Production AI at scale behind it. Voice agents handling 500+ concurrent calls, RAG over terabytes of documents, LLM quality checks over millions of support chats.
- The same rule for clients and for me. Mainstream, vendor-supported tools. If a step would need custom code, I say so, and usually recommend not doing it.
Measured results, not AI theater.
A few outcomes from production systems I designed and led. The full story is in the resume.
Enterprise RAG Challenge 3, against industry teams.
concurrent calls handled by a real-time AI voice agent for marketing campaigns.
RedCore (former PIN-UP)support-chat QA coverage with LLM checks; assessment accuracy rose from 66% to 91%.
RedCore (former PIN-UP)to find information in a terabyte-scale RAG platform on AWS Kendra and Bedrock.
RedCore (former PIN-UP)for an AI voice ordering agent with a 91% completion rate.
Milina, New York hospitalityafter rebuilding an LLM-agent platform; client bots built 40× faster.
GartTechAlso: appointed expert to the European Commission working group TWG6 (Women Shaping the AI Economy), technical reviewer for Manning Publications, Pro member of AI Chamber CEE.
Where I share what I learn.
Profiles
Guides and e-books
Say hello.
For an AI operating system for your business, start at aiinorder.com. For anything else (partnerships, products, speaking, a role), write to me directly.
- Telegram
- @vandriichuk
- linkedin.com/in/viktorandriichuk
- AI in Order projects
- hello@aiinorder.com