Welcome to CTO eXchange!
Topic: From Chat to Action: Building Real Enterprise Knowledge Systems with LLMs
Enterprises sit on vast amounts of information spread across documents, systems, and formats that were never designed to work together. Large Language Models promise a shortcut—but on their own, they know nothing about your domain, your data, or your constraints.
In this talk, Michal Stypa and Johan Henriksson share key learnings from building enterprise-grade knowledge retrieval systems using LLMs, agents, and retrieval-augmented generation (RAG). We explore how context is constructed, how information is retrieved and summarized, and why naïve “chat with your documents” approaches quickly break down in real-world environments.
The focus is on practical, hands-on lessons from designing agentic knowledge retrieval systems on top of enterprise data: where retrieval works, where it fails, and how these systems evolve from answering questions to driving automated workflows. Chat is a temporary interface—the real value lies in eliminating manual work.
This session is aimed at technical practitioners who are already implementing, or planning to implement, document-based knowledge retrieval systems in an enterprise setting.
🗓 Date: 24 February
🕒 Time: 11.30-13.30
📍 Location: The Tribe Malmö, Skeppsbron 5
Take the opportunity to network with other CTOs, exchange experiences and get concrete tips to take your team to the next level.Register now and secure your place!
You will also meet a representative of CTO eXchange; Andreas Jangefalk and today's Sponsor Developers Bay; Jenny Andersson and The Tribe; Jenny Fergéus Almroth
Target group: You work as a CTO, Development Manager or in a similar role
What you can expect
At CTO eXchange, the participants have the central role and can get help from the expertise and experiences of others but also assist with good advice and experiences themselves. Each opportunity will have different topics and focus areas to discuss together.
After introduction, we will be divided into smaller groups and together discuss some of the issues we will look at together, but not limited to the following:
Questions:
1. Which documents or knowledge domains do you use today, or see a clear use for, in an internal chat or Q&A setup and what problem are you actually trying to solve?
2. Which risks are you most concerned about when exposing an LLM-based system to your internal data?
3. Which manual, knowledge-intensive process would you most like to eliminate entirely, if the technology was reliable enough?
You are welcome to contribute more suggestions for questions via the registration link.
Lunch is served 11.30 - register allergies when you sign up.
We hope you will enjoy the session—and come back for future events.
