- Dates: November 17 to December 1, 2026
- Fee: €360
- Number of places: 10
- Application deadline: Midnight CEST on November 15, 2026 (or until the course is full)
- Questions? Contact Ada: ada@journalismarena.eu
Join our hands-on masterclass for journalists who want to learn how to use local coding agents.
We’re all questioning AI right now: should we be using it; what can and can’t models do; how do we check its work against the source data, knowing when, and when not to trust the output?
Have you ever…
- Spent half a day manually cleaning a spreadsheet, and another half checking whether the results are actually useful?
- Asked ChatGPT to analyse a dataset, received an answer that sounded plausible, but wondered if it interpret the data correctly?
- Tried using AI in your workflow already, but found yourself stuck in the chat box?
In this masterclass we will be focusing on developing workflows using AI agents locally on our laptops using free open source tools: opencode.ai and Obsidian.
You’ll walk away with:
- Practical command of AI agents for coding, data analysis, and research,
- Ready-to-use workflows to take into the newsroom.
This course is for:
- Reporters, editors, and researchers with some experience in data analysis: you should have looked at a dataset before and know what a pivot table is.
- No coding experience is required. You’ll need a laptop, dedication, and curiosity.
Important note: local agents mean you will give the model access to your computer. It will be able to read, write and execute code on your local machine. It does not mean we will be using local models. You can work with your own AI subscriptions, or we will provide you with a token budget to work with for the scope of the course.
Our track record
As experienced journalism trainers, and the parent company of the Dataharvest conference, Arena for Journalism in Europe has been supporting data journalists to investigate, collaborate and network for more than 15 years. Journalists who have received our training have gone on to publish agenda-setting investigations and win accolades including Sigma Awards, the IJ4EU Impact Award and the European Press Prize.
Many people find life-long collaborators on our programmes and we are proud to continue to support them with mentoring, IT, networks, conferences, research and more.
Examples of data investigations published following Arena training:
- Example
- Example
- Example
Ada Homolova is a freelance data journalist with over a decade of experience with data storytelling. She has crunched data for among others OCCRP, Follow The Money, Correctiv and Lost in Europe and coordinated the skills track of the Dataharvest conference since 2020. Last year she started her own knowledge sharing platform Data Frosch.
Course structure
- There are 10 places on this course; a small cohort so everyone’s questions get answered.
- There are 3 live, 2-hour sessions on Tuesday mornings each week, starting November 17.
- Each session will provide homework with a fresh dataset you can practice with; in total you should allow for 2 hours each week to attend the sessions and additional 2 hours to complete the homework.
All times in CEST
Nov 17, 2026 |SESSION 01: Robots in your laptop
10am – 12pm | Beyond a chat window: an agent that reads your files, runs code, and shows its work.
Take a small dataset, find a simple story, and produce a chart. In the meantime we’ll learn:
- What a local coding agent actually is: the loop it runs, the files it can touch
- Currently available models and their pro’s and cons
- Costs, limits, and data protection: what leaves your machine, what stays local, and what that means for sensitive source material
- Context building: keeping your agent on track
+ Homework
Nov 24, 2026 | SESSION 02: Research & analysis with LLMs
10am – 12pm | Stop copy-pasting prompts from the internet.
A larger data flow: cleaning, merging, summary stats + creating an exploratory dashboard. In the meantime we’ll learn:
- How to use sub-agents
- The power of automatic research
- How to automate and fact-check cleaning, merging, summary statistic, automated testing
- Exploratory dashboards as a way of sharing the analysis
+ Homework
Dec 1, 2026 | SESSION 03:Coding with LLMs
10am – 12pm | The data you need isn’t always the data you’re given. Go get it.
Write a Python scraper with LLM assistance. In the meantime we’ll learn:
- What a web page actually is, and reading its structure with the browser inspector
- The Network tab: where undocumented APIs and hidden data live
- Scraping responsibly: rate limits, terms of service, robots.txt
- Getting away with not being able to code
+ Homework
Fees
A full place on the course costs €350.
- Included: module tuition, support in between sessions in a dedicated group chat.
- Discounts: A number of discounted places are available for those for whom fees are a barrier — if you need a discount please write a brief motivation to the course leader Ada Homolova directly.
Deadline
Applications close at midnight CEST on November 15, 2026 (or until the course is full). Submit your application via this form.