How To Submit A Conference Talk
Your next AI conference talk starts with a problem you’ve already solved
The AI prototype that worked beautifully until someone uploaded an unexpected document. The evaluation that made you question your original approach. The workflow your team redesigned after discovering where human review was still essential.
These experiences can feel like ordinary parts of working with AI. They can also be the starting point for a conference talk that helps someone else make a better decision.
If you have considered speaking at an AI conference but struggled to find a topic, look at the work already on your desk. What have you investigated, changed or learned that another team could benefit from?
Start with the question that slowed you down
Think back to a moment when the next step wasn’t obvious.
Perhaps you were choosing between models, investigating unreliable answers or deciding how to evaluate an AI feature before releasing it. Maybe the difficult part was understanding what users actually needed, or agreeing on when the system should ask a person for help.
What made the question difficult? Which options did you explore? What evidence helped you decide?
Answering those questions gives your talk a clear direction. It also helps you identify the audience: people facing a similar decision, with similar constraints.
Your implementation choices are worth explaining
You don’t need to have developed a new model to contribute a useful perspective on applied AI.
There is plenty to explore in how you use existing technology. Consider the decisions behind a document search tool, an AI coding workflow or a machine learning pipeline. How did you choose your approach? What did you measure? Where did cost, latency, data quality or maintainability influence the outcome?
For example, if you worked on a retrieval-augmented generation (RAG) application, you could focus on how you tested whether it retrieved the right information. Show the questions you used, what the results revealed and which changes you made.
That gives the audience something concrete to examine and adapt to their own work.
Share what changed your mind
A project that took an unexpected turn can make a compelling technical story.
Perhaps your initial evaluation looked promising, but feedback from users revealed a different problem. Maybe an AI agent performed well on individual tasks but needed clearer boundaries in a longer workflow. Or your team discovered that a simpler approach met the requirements.
Explain your starting assumptions and the evidence that challenged them. Be clear about what you observed, what you measured and what remains uncertain.
You don’t need a perfect ending. An honest account of what you would test differently next time can be a useful takeaway in itself.
Turn a broad subject into a focused talk
If you are looking for AI conference talk ideas, try narrowing a familiar subject to one practical question:
- LLM evaluation: How did you build a test set that reflected the questions your users actually asked?
- AI agents: Where did you introduce human approval, and how did you decide?
- AI-assisted development: How did your team review generated code and assess whether it helped?
- Machine learning operations: How did you investigate a change in model performance after deployment?
- Human–AI interaction: What did you change when users misunderstood an AI feature’s limitations?
Choose a question you can support with examples. A focused session gives you space to explain the details, including the approaches that didn’t work.
How to submit an AI conference talk
Once you have a topic, write down three things: who the session is for, which problem it addresses and what attendees will learn.
Use those answers to shape your abstract. Explain what you will cover and why it matters to that audience. Include enough context for reviewers to understand your contribution, and make the expected knowledge level clear.
Before submitting, ask a colleague to read the proposal. Can they explain what they would learn from attending? If their answer is vague, make the scope or takeaways more specific.
Check which examples you can share publicly, and prepare anonymised or synthetic material where needed.
Share your experience at Dutch AI Conference 2027
At Dutch AI Conference, we welcome practical insights into building, applying and working with AI. That includes technical deep dives, architectural decisions, responsible AI and the ways AI changes development workflows and teams.
Whether you are preparing your first proposal or returning with a new topic, start with an experience you can explain clearly.
What challenged you? What did you try? What could someone else learn from the result?
The Dutch AI Conference 2027 Call for Papers closes on 18 December 2026.
(For full guidelines, visit https://aiconference.nl/meet-the-speakers/call-for-presentations/).
