Speakers
Lucas de Araújo
Ethira - Startup Founder
Speedrunning a startup with AI-driven engineering
Lucas will share insights into Ethira's journey as a startup building AI for risk management, fully utilizing AI in their coding from the very beginning.
Hear from a startup leveraging AI-assisted development from day one, the challenges faced, and the benefits gained from integrating AI into every aspect of the development process.
- Secure software development with AI
- Productivity hacks
- Code hygiene
Göran Krampe
Trustly
How an Agentic Coding Assistant Works
Göran dives into the internals of agentic AI coding assistants – what's actually happening under the hood? This technical exploration will give a taste of the LLM protocol used by coding assistants and highlight different challenges in implementing agentic tools like Claude Code.
Having an understanding of how the tools work makes it easier for us to know how to use them. This is an area of frantic innovation and there is now an abundance of tools available to us developers.
- OpenAI API
- System prompt composition
- The agentic loop and context management
- Sub agents and tool calling
- Code base indexing
Daniel Trugman
Requesty.ai - Co-Founder
LLM Protocols and LLM Routers
When large language models move into production, reliability and cost efficiency quickly become central concerns. This talk looks at the practical infrastructure layer that makes LLM systems dependable and affordable. We will explore how routing and failover between different model providers actually work, how to handle provider-specific error codes, and how to design systems that continue to function when models or endpoints fail.
The session also covers caching and request optimization strategies that help reduce costs without sacrificing performance, and the trade-offs between managed SaaS solutions and self-contained environments like Bedrock.
- Model routing and failover mechanisms
- Handling provider-specific error codes and inconsistencies
- Designing reliable and observable LLM agents
- Caching and request optimization to manage cost
Bishal Thapa, Caio Silva, Mikko Ekström, Abdulkadir Olmez, Fredrik Duong
Trustly
Report from the AI Pilot
Our internal team shares their experiences and findings from Trustly's AI pilot program – what worked, what didn't, and lessons learned along the way.
Get a series of real demonstrations of AI tool adoption, and recommendations for teams going forward.
- Mikko Ekström: Bridging the Terraform Gap: From Idea to MVP with AI and MCP
- Abdulkadir Olmez: Efficiency and Quality Improvement in C# Project Analysis with Modular AI Instruction Sets
- Caio César Silva: AI-Powered Order Investigation: Production Log Analysis with Claude
- Bishal Thapa: Supercharging Local Development Building AI-Driven Internal Tooling
- Fredrik Duong: AI to the Rescue: Replacing External Dependencies in Qlik Cloud with Cursor
Mattias Jonsson
AI, Scale, HA and Databases
AI solutions are composed of many different components, and data handling is a center part of it. The session will cover several components, like Large Language Models (LLM), Semantic Search, Vector embeddings, Retrieval Augmented Generation (RAG), Model Context Protocol (MCP), and how a database can be serving the different components.
The session will also address challenges of High Availability and Scalability, using TiDB's architecture as an example how a modern database meets these demands.
- AI components explained, LLMs, Vector embeddings, RAG, MCP
- Database requirements for serving AI agents
- HA and Scaling of a modern Distributed SQL database, TiDB
Jonas Lindskog
Trustly
What Happens Now?
Jonas tries to connect the dots on what happens when AI takes a major part in software development. This closing talk will synthesize the day's insights and learnings and look toward the future.
What are the strategic implications of AI-assisted development. What does it mean for me as an engineer and what does it mean for Trustly as a company? Hopefully this session will leave you will some answers to all the questions you have.