I built an entire AI-run company. Solo.
2 MCP servers · 27 AI agents · 32 microservices · 2 SaaS products — one engineer. 16+ years full-stack. Now I build agentic AI that actually runs in production for your team.
Brands I've built for
AI Operations, in production
- 2 MCP servers + pgvector hybrid search
- 27 role-specialized agents shipped
- GDPR / DSGVO by design
186+
lead-gen sites running
A complete, AI-run operating system — designed, built and shipped to production by one engineer.
27
AI agents in production
32
microservices shipped solo
49
integrations wired up
Agentic AI that runs real operations
Three things I do, end to end — and ship. The stack is proof, not the headline: the headline is what your team gets back.
- Agentic AI & automation — multi-agent systems on MCP that run CRM, tasks, outreach and content.
- Full-stack delivery — TypeScript/Node, Vue/Nuxt, Postgres/pgvector, shipped to prod, not notebooks.
- Growth engineering — programmatic SEO and AI lead-gen at scale (~186 sites running).
Automation throughput
tasks closed by agents
See the systems actually working
Not slides — live systems I built and run in production. Open a case to see exactly how each one works.
Control Center — agents moving tasks
A single ops cockpit: 27 agents pull work, ship outcomes, report stats.
AI-Souffleur — real-time call assistant
German calls: mic → STT → Claude prompts → TTS, live in the ear.
Tender engine — cross-portal bid pipeline
15+ procurement portals → one DB → semantic match → PDF/DOCX autofill.
Find your situation — and what I build for it
Lead-gen agencies & marketers
Manual lead handling, leaky funnels → AI funnels, quiz flows, ~186-site programmatic SEO, CRM & follow-up automation.
Local & service SMBs
Missed calls, slow follow-up → real-time call assistant, inbound capture, Google-Maps/local-SEO pipeline, email engine.
Public-tender bidders
Tenders across 15+ portals, manual bids → cross-portal aggregation, semantic match, PDF/DOCX autofill.
Product & SaaS teams
Want AI features but never reach prod → LLM/agent in product, MCP/tool-use, RAG, monitoring + cost control.
Finance / insurance / advisory
Leaking leads, manual nurture, compliance → email funnel engine, Pipedrive CRM, GDPR-by-design portals.
Ops-heavy companies
Routine: invoices, reports, data entry, meetings → multi-agent ops, document automation, cashflow, meeting→tasks.
Didn't find your case? Tell me on a call — if it’s routine, it can be automated.
Systems I designed and shipped
Each one is real, in production, and built solo.
Brainbase MCP
Self-built MCP server, single entry, 5 access scopes.
Memorium v2
Org-memory: Postgres + pgvector hybrid search (vector+BM25+RRF).
Multi-agent platform
27 role-specialized agents on Claude/GPT via one LLM gateway.
AI-Souffleur
Real-time German call assistant: STT → Claude → TTS.
GDPR-by-design CRM
Headless CRM with Brevo inbound parsing + DSGVO export.
Quiz Platform
Serverless multi-step lead-gen quizzes, JSON-configured.
linkobox
Productized SaaS: link-in-bio profiles + KI-avatar outreach video.
Landing Forge
SaaS: one Nuxt codebase → N niche landing pages from JSON.
Systems I shipped end to end
The tools behind the systems
AI / Agents
Engineering
Integrations
Growth
How one engineer built all this
I'm Alex — 16+ years shipping full-stack software (TypeScript, Node, Vue/Nuxt, Postgres). In 2024 I went all-in on agentic AI and, solo, built an entire AI-run operating system for a real services business: 2 MCP servers, 27 agents, 32 microservices, 2 SaaS products. Not prototypes — production, with monitoring, GDPR and cost control.
Now I build that capability for other teams — fast, hands-on, and without needing to be managed.
16+
years full-stack
2024→
all-in on agentic AI
DE
Munich-based, EN/DE/RU/UA
Alex Grygoriev
Senior AI Automation Engineer · München
Pick how we start
No risk, no fluff — every track has a guarantee. Price is set on the call.
AI Ops Audit
For teams drowning in manual work
A map of where AI takes 10–20 hrs/week off your plate.
What's included
- Process audit
- ROI-prioritized automation map
- 1 quick-win demo
2-Week Agent Pilot
flagshipFor those who want a result, not a deck
A working AI agent for one of your processes — in prod, in 14 days.
What's included
- Working prototype
- Integrated into your stack
- Handover & docs
LLM Feature in your product
For product & SaaS teams
Your product gets an AI feature that runs in prod — monitored, cost-controlled.
What's included
- LLM/agent integration (MCP/tool-use)
- Eval + cost control
- GDPR-aware
Hiring? Not sure? Start with a paid trial project / contract-to-hire — zero risk for your headcount.
Frequently asked questions
A working prototype in ~2 weeks, in production. I’ve shipped 32 microservices solo — I move fast and I ship, not just prototype.
Let's build something that runs.
Pick a slot in my calendar, or send a message. One click, no friction.
Book an intro call
Pick a free slot below — instant confirmation, no back-and-forth.
Opens an embedded Google Calendar. By clicking you consent to loading it (see Privacy).