Show the room what works, and where it breaks

Our audience has seen enough demos that only work on stage. They come to learn from people who ran the thing in production, hit the wall, and can say what they would do differently. That is where you come in.

You do not need a groundbreaking idea, and you do not need to be a seasoned speaker. A problem you solved, a migration that went sideways, or a design decision you would defend (or reverse) makes a better session than a tour of new features. We read every submission, and if you want a second pair of eyes on an idea before you submit, just write to us.

Ways to speak

• Main conference session, 20 May at CUBEX. 50 minutes including 5-10 minutes of Q&A. Parallel tracks, all levels, weighted toward depth. We plan around 40 sessions.
• Lightning talk, 20 May at CUBEX. 20 minutes, one idea, one takeaway. A good first step if you have not spoken at a conference yet.
• Pre-conference workshop, 19 May at VŠE. A full day (6 hours), hands-on, with a clear skill attendees take home. We select 6. Workshop speakers get a revenue-sharing model, plus travel and accommodation covered. If you co-present, agree on the split among yourselves before you submit.
• Panel discussion. Propose a topic and the people for it.

What we are looking for

We program by theme. If your idea fits one of these, tell us which one. If it sits between two, that is fine too.

• Agentic analytics and data agents. Fabric data agents, Genie, Cortex, Copilot Studio, Microsoft Foundry and others in real use, including where agents fail.
• Semantic layer, ontology and context. Semantic models, metrics and metrics layers such as the dbt Semantic Layer, Fabric IQ and ontologies, open semantic standards, and how to give an agent the context it needs.
• AI-assisted development. Coding agents and assistants in data work, MCP servers, agentic engineering, TMDL and notebooks written with AI.
• Data engineering, pipelines and analytics engineering. Ingestion, orchestration, Spark and SQL, dbt, Python and Polars, DuckDB, testing and reliability.
• Lakehouse, warehouse and open formats. OneLake, Delta, Iceberg, mirroring, Databricks, Snowflake, and how they work together.
• Real-time, operational data and Postgres. Real-Time Intelligence, streaming, SQL database in Fabric, PostgreSQL.
• SQL Server, Azure SQL and database craft. Performance, internals, administration, HA/DR, modernization.
• Power BI, DAX and modeling craft. Direct Lake, DAX, translytical task flows, TMDL and PBIP, report design.
• Governance, security and AI compliance. Purview, OneLake security, data quality, observability, EU AI Act in practice.
• Modernization, DataOps and FinOps. Migrations (Synapse, ADF, SSIS, on-prem SQL Server), CI/CD, Git, capacity and cost control.
• People, career and community. Non-technical sessions are welcome and valued.

On AI tools: we are vendor-neutral. Sessions about Claude, ChatGPT and Codex, Microsoft Copilot and Foundry, GitHub Copilot, Cursor, Databricks, Snowflake, and open-source or local models are all welcome.

Tell us what you used, what it cost, and what went wrong.

What helps your session stand out

• A clear audience and a clear takeaway: who should be in the room, and what can they do on Monday that they could not do on Friday?
• Evidence from real work: a real number, a real architecture, a demo that is allowed to fail.
• An honest level: 100 introductory, 200 intermediate, 300 experienced, 400 advanced or expert. For 400 please list the prerequisites, so attendees can choose well.
• An abstract written by you. Using AI to polish it is fine. You own every claim in it, and we may ask to see how the session holds up.

What is not a good fit

• Product pitches. Vendor engineers and partners are very welcome when the session stays useful without the logo, ideally told together with a customer.
• A walk through release notes or documentation, which the audience can read themselves.
• Demos without architecture, or sessions that do not say what you would do differently.

In person only

Data Point Prague is an in-person event, so speakers travel to Prague. We do not run remote or hybrid sessions: the room is a big part of what makes it work, and we want you in it.

What you get

• Free event ticket for all accepted speakers
• One night of accommodation at Hotel Panorama, right next to CUBEX
• Pre-conference workshop speakers: travel and all nights of accommodation covered, plus revenue sharing
• Speaker dinner and side events with the organizers and the other speakers

Timeline

• Call opens: 16 October 2026
• Call closes: 31 January 2027 (23:59, Prague time)
• Decisions: within 14 days after the call closes
• You can submit up to 4 sessions. Pick your best ones.

Questions before you submit

Not sure whether your idea fits? Send us two sentences and we will give you honest feedback and help you shape it. Write to Jiri at jiri.starostik@databrothers.cz.

One night of accommodation at Hotel Panorama, right next to CUBEX. Pre-conference workshop speakers: travel and all nights of accommodation covered.