Lukáš Orčík

I Build AI Agents for Businesses

Results, not presentations.

20 years in IT, the last 5 exclusively with AI. I develop agents, browser automation, and multi-agent systems — from the first prototype to production deployment.

Why people reach out

Systems that hold

Production deployment without downtime — even under heavy load

Fast start

Working PoC in 2 weeks, so you know what to expect

I understand business

I want it to make money — not just look good

About Me

Most companies today want AI. The problem isn't technology — it's available. The problem is knowing exactly where to start and what to do first.

I have research from VŠB-TUO and years of Big Data work in telecommunications behind me. I understand what's happening under the hood of LLMs — and know how to apply it to a real business problem, not just a demo.

Technology for its own sake doesn't interest me. I want what I build to earn or save your company money — as soon as possible.

Why me?

  • Research backgroundI know what's happening inside the models — I don't sell black boxes.
  • Data experienceYears of working with large and sensitive data in a telco environment.
  • No server worriesCloud-native approach — I deploy, monitor, scale. You focus on business.

By the numbers

Years in IT

20+

Completed projects

30+

Satisfied clients

30+

What I do

Four things companies need most and pay for with manual labor

  • 01

    AI agents instead of manual repetition

    Data entry, approvals, checks, monitoring — things you repeat every day. I replace them with an agent that works unsupervised, but with your eye on the critical points.

    • LangGraph orchestration
    • Supervisor pattern
    • Human oversight where it matters
    • Complete audit trail
  • 02

    AI connected to your systems

    CRM, ERP, databases, company documents — AI is useless without access to them. Model Context Protocol (MCP) works as a connector and saves a lot of integration work.

    • Custom MCP servers
    • Agentic RAG
    • PydanticAI data validation
    • Private LLM (DeepSeek)
  • 03

    Scraping without regular maintenance

    Classic scrapers break with every website redesign. The agents I build see the page like a human — no CSS selectors, no code rewriting when layouts change.

    • Browser-Use + computer vision
    • Self-healing logic
    • No HTML structure dependency
    • Scaling to thousands of URLs
  • 04

    EU AI Act: what exactly you need to comply with

    If you're deploying AI in HR, customer support, or decision-making, the EU AI Act applies to you. I'll help figure out what exactly you need to do, and we'll go through it step by step.

    • Readiness audit
    • System risk classification
    • Employee training
    • Documentation and reporting

Skills & Technologies

Overview of technologies and tools I work with daily

AI Agents & Orchestration

LangGraphPydanticAIAutoGenBrowser-UseMCP ProtocolAgentic RAGHuman-in-the-Loop

LLM Models

Claude 4 SonnetDeepSeek R1/V3GPT-o3Gemini 2.5OpenAI APIAnthropic API

Cloud & Infrastructure

AWS BedrockLambdaFargateS3EC2vLLMCloudNative

DevOps & Automation

DockerKubernetesTerraformCI/CDGitHub ActionsAnsible

Programming

PythonTypeScriptJavaScriptGoBashSQL

Big Data & Databases

PostgreSQLMongoDBRedisElasticsearchPineconeETL

Web Development

ReactNext.jsNode.jsTailwind CSSREST APIGraphQL

Case Studies

Real results I've achieved for my clients

Finance & LawAudit Group

Autonomous Accounting Agent

Challenge:

Monitoring constant legislative changes and manually checking thousands of documents against new laws.

Solution:

Multi-agent system with LangGraph orchestration and Supervisor pattern. Nonstop legislative monitoring with automatic document validation and Human-in-the-Loop approval.

98%
compliance
120h
time saved/mo.
Real-time
monitoring
LangGraphPydanticAIClaude Opus 4.5+1 more
Retail & MarketingE-commerce Platform

AI Market Intelligence Scraper

Challenge:

Need to monitor competitor prices and trends across thousands of websites without constantly writing new scrapers.

Solution:

Autonomous scraper using the Browser-Use model, which 'sees' the web and extracts data regardless of changes in site code.

-90%
maintenance cost
5x
more data monthly
2 weeks
deployment
Browser-UseGemini 2.5 FlashPlaywright+1 more
Compliance & LegalInternational Corporation

Legislative Guard Agent

Challenge:

Complex process of comparing internal guidelines with new EU regulations.

Solution:

Agentic RAG with DeepSeek R1 for deep legal reasoning. Agent analyzes conflicts between new legislation and internal policies, generates Gap Analysis and proposes specific text changes.

90%
risk accuracy
60%
faster revision
24/7
analysis
LangGraphDeepSeek R1Claude Opus 4.5+1 more

Most common questions

Things people ask me before they write.

Wondering what could be automated?

Write to me about what's bothering you or what you want to automate. In the first (free) consultation, we'll figure out if and how AI can help — no commitments, no buzzwords.

Free first consultation
PoC within 2 weeks

Contact Me

Have a project or question? Don't hesitate to reach out.

Location

Olomouc

I respond within 24 hours

Social

Orcik Tech s.r.o.

ICO: 08763941|VAT: CZ08763941

Jakoubka ze Stříbra 781/44, Nové Sady, 779 00 Olomouc

Registered at: Krajský soud v Ostravě, C 80619/KSOS