Date: Tuesday, 07 July 2026
Schedule: 10:00 – 14:45
Location: Datalab Meeting Room / Online
Purpose
Showcase the best use cases of AI in programming. Participants are encouraged to present their own best practices if they have something to share.
Challenge
The most impressive examples will be rewarded.
Language
English, Bosnian, Serbian, Croatian
Pizza and drinks provided by HQ.
Participants
- HQ Programmers
- SUB Programmers
- PM/APM
10:00 SUIV team
AI Security & Best Practices
Lecturer: SUIV team
Duration: 30 min
Discuss security, privacy, and intellectual property considerations when using AI tools in software development. Explain what types of source code, project information, and customer data can be safely shared with AI tools, what should be avoided, and which best practices developers should follow to use AI responsibly and in compliance with company policies.
Topics:
- Which code can be shared with AI tools?
- What information should never be shared?
- Customer and personal data considerations
- Intellectual property and source code ownership
- Public vs. enterprise AI tools
- Practical guidelines for safe AI usage
- Real examples and common mistakes
- Q&A
Outcomes:
- what they can paste into ChatGPT/Claude,
- what they are not allowed to paste,
- how to anonymize data,
- when to use the enterprise version,
- and what rules apply within Datalab.
10:30 Dušan Majkić
AI-Assisted Development in the PAAS Project
Duration: ~30 min
Presenter: Dušan Majkić, Datalab Tehnologije d.d.
AI tool used: Claude (Claude Code)
I will present how I use AI assistance when experimenting with the PAAS project — a compact and approachable codebase that is ideal for trying out new development approaches. Through practical examples, I will show how Claude helps in understanding existing functionality, identifying issues, implementing new features, and improving overall development efficiency.
Topics
- Introduction to AI-assisted development in PAAS
- Overview of Claude Code and how it is used in practice
- Understanding existing code and project structure with AI support
- Faster implementation of new features using AI assistance
- Debugging and problem-solving with Claude
- Improving code quality and refactoring with AI
- Key lessons learned and best practices
- Q&A
11:00 Aljoša Žvokelj
AI-Assisted Development of Real-Time Speech Recognition
Duration: ~20 min
Presenter: Aljoša Žvokelj, Datalab Tehnologije d.d.
AI tool used: Claude
How Claude helped me design, code, debug, and refactor a complete real-time speech-to-text system. From architecture whiteboarding to “why is my text in the wrong field?” to translation refactoring. Iterative development, clear process, and why AI excels at each step.
Topics
- Understanding Soniox platform — APIs, documentation
- Architecture Design — eliminating bad ideas before coding
- Debugging — Claude’s Socratic questions
- Refactoring & Translation — Claude approach
- Iterative Development — 3-4 rounds per feature is normal
- The Workflow That Worked — Paste error → clarifying questions → context → reason through invariants → iterate
11:20 Coffee break
11:40 Marko Tuk
AI-Assisted SQL Development with SSMS Copilot
Duration: 30 min
Presenter: Marko Tuk, Datalab Tehnologije d.d.
AI tool used: SSMS Copilot
I will present how SSMS Copilot can support SQL Server development and DBA work directly in SSMS. Through practical examples, I will show how it helps generate stored procedure templates, convert business requirements into SQL, rewrite cursor logic as set-based operations, analyze and optimize existing procedures and queries, suggest indexes, refactor legacy code, generate CRUD procedures, and assist with execution plans and administration tasks.
Topics
- Introduction to AI-assisted SQL Server development in SSMS
- Generating stored procedure templates with TRY/CATCH, transaction handling, logging, and return codes
- Converting business requirements into SQL procedures
- Rewriting cursors into set-based SQL operations
- Analyzing, reviewing, and refactoring existing stored procedures
- Optimizing slow queries, execution plans, and index suggestions
- Generating CRUD procedures and finding dependencies between database objects
- DBA use cases: blocking sessions, long-running transactions, largest tables, and unused indexes
12:10 Andrej Marinič
AI-Assisted Development: From Legacy Code Migration to Automated Testing
Duration: 45–60 min
Presenter: Andrej Marinič
AI tool used: ChatGPT / Claude
PART 1: Delphi to .NET Migration with AI Assistance
I will present how AI can accelerate the migration of legacy Delphi applications to modern .NET solutions. Through a practical migration example, I will demonstrate how AI helps analyze Delphi units, identify business logic, understand dependencies, design a maintainable .NET architecture, and generate implementation guidelines that can be reused in future migration projects.
Topics:
- Challenges of migrating legacy Delphi applications
- Understanding business logic with AI assistance
- Extracting dependencies and architecture patterns
- Converting Delphi concepts into .NET services and projects
- Practical migration example (Delphi → C#)
- Building reusable migration documentation and standards
- Lessons learned and recommendations for future migrations
PART 2: From Requirements to SQL Solution Using AI
I will demonstrate how AI can transform a simple business requirement into production-ready SQL code and documentation within minutes. Using a workflow parameter loading example, I will show how iterative communication with AI helps developers refine requirements, improve code quality, and create reusable implementation patterns.
Topics:
- Converting business requirements into technical solutions
- AI-assisted SQL procedure development
- Iterative refinement of generated code
- Using CTEs and ranking logic for configurable workflows
- Creating reusable database patterns
- Generating technical documentation automatically
- Improving developer productivity with AI
PART 3: From Source Code to Tester Scenarios
I will show how AI can help transform complex source code into structured tester documentation and test scenarios. Through a practical Delphi example, I will demonstrate how authorization checks, business rules, and user interactions can be automatically converted into understandable testing procedures and visual documentation.
Topics:
- Analyzing existing source code with AI
- Identifying business rules and decision branches
- Transforming code into tester-friendly scenarios
- Generating test cases from application behavior
- Creating process diagrams and supporting documentation
- Improving collaboration between developers and testers
- Building repeatable workflows for test preparation
Q&A
12:40 Lunch
Pizza and drinks for live participants
13:40 Bojan Lenarčič
AI-Assisted SQL function
Duration: ~3 min
Presenter: Bojan Lenarčič, Datalab Tehnologije d.d.
AI tool used: Chat GPT
Sample function fHR_GetWorkerHierarchyOnDate.
Topics
- controls that it does not cycle using the path
13:45 Gašper Petelin
Use of AI-assistant to develop for non-developers
Duration: 15 min
Presenter: Gašper Petelin, Datalab d.d.
AI tool used: Microsoft Copilot
I will present how AI assistant was used by non-developer to create procedure for Themis. Presentation shows steps prior to development and development process itself. At the end it includes review of the process and identifies improvement proposals to continue with such work.
Note: presentation is intended for non-developers, not so much for programmers.
Topics
- Description of the challenge and used tools
- Preparation before development
- Development process
- Review of results
- Improvement proposals
- Q&A (via mail or Teams)
14:00 Aleš Golob
Converting Delphi objects to MS SQL code
Duration: 15 min
Presenter: , Aleš Golob, Datalab Tehnologije d.d.
AI tool used: Gemini
I will present how I use the help of artificial intelligence to convert a Delphi function or procedure into an MS SQL function or procedure.
Topics
- Converting a Pascal procedure to an MS SQL stored procedure.
- Converting a Pascal function to an MS SQL function.
- Converting a MS SQL scalar function to an MS SQL ITF function.
14:15 Andrej Gerbec, eVizija
AI Code Review Automation with Local LLMs in PAW
Duration: ~30 min
Presenter: Andrej Gerbec, E.Vizija d.o.o.
AI tool used: Ollama (Qwen3-Coder), GitLab CI/CD, ChatGPT
I will present how we use AI assistance in a real software development workflow by integrating local AI models into our code review process. Through practical examples, I will show how AI helps developers analyze merge requests, detect potential issues, improve code quality, follow development standards, and increase team efficiency while keeping source code inside the company environment, aiming to remove the waiting periods for human reviews, which are increasingly proving to be the bottleneck.
Topics
- Why we introduced automated AI code reviews
- Overview of the architecture:
- GitLab CI/CD pipeline
- local Ollama server
- Qwen3-Coder model
- automated merge request analysis
- Reviewing real-world C#/.NET and Angular applications with AI
- Detecting common development issues:
- security vulnerabilities (OWASP, SQL injection, XSS)
- bad async patterns and performance problems
- architecture and coding standard violations
- maintainability issues
- Creating effective AI prompts and review guidelines
- Practical examples: what AI found in real code
- Challenges and lessons learned:
- model selection
- context size limitations
- false positives
- balancing AI and human review
- Future ideas and opportunities:
- automated test generation
- documentation assistance
- AI-assisted development workflows
- Reverse Q&A:
- gathering ideas, experiences, and feedback from the audience
- discussing real-world challenges others are facing with AI adoption
- exploring where AI can provide the most value in development teams
14:45 End of Event
X
Datalab Tehnologije d.d.
Hajdrihova ulica 28c
1000 Ljubljana
Slovenia
