The Cochrane Colloquium 2026 takes place at ICE Kraków from 8 to 10 December. For Evidence Prime, the team behind Laser AI and GRADEpro, this year's Colloquium is a home event: we are based in Kraków, and Artur Nowak is a member of the Local Organising Committee. He also reviewed abstracts submitted to the Colloquium.
Our contributions span the week, from GRADE methods and the responsible use of AI in evidence synthesis to moving evidence into guideline recommendations. This page lists where you can find us, and we will update it as the details are confirmed.
Before the Colloquium
Sunday 6 December: GRADE Working Group meeting. Members of our team will join the GRADE Working Group meeting held in Kraków ahead of the Colloquium. As the developers of GRADEpro, we follow the group's methodological work closely, because new guidance eventually needs to be reflected in the tools that guideline panels use.
Monday 7 December: Methods Symposium and ICASR meeting. This year, the annual meeting of the International Collaboration for the Automation of Systematic Reviews (ICASR) takes the form of a joint Methods Symposium on advancing methods for the responsible use of AI in evidence synthesis, co-hosted with the Cochrane Methods Executive. Evidence Prime has been part of ICASR for many years, and Artur Nowak has taken part in its meetings since 2016. As a collaboration, ICASR builds its meetings around shared work rather than a series of talks, and the symposium follows that approach: expert presentations are combined with structured small-group discussions. Artur will join this work, which focuses on how AI tools for review tasks should be evaluated and what evidence teams need before adopting them.
Oral presentations
On Tuesday 8 December, Artur Nowak will present two studies from our team in the session Responsible AI in evidence synthesis 1 (ICE Kraków, S2 Theatre).
- 11:55–12:10: Give me instructions: experimental evaluation of different ways of prompting large language models for artificial intelligence-assisted title and abstract screening. The study compares published eligibility criteria used verbatim with decision rules written by the team and decision rules generated by a model, across seven published reviews. Authors: Ewa Borowiack, Monika Opalek, Artur Nowak and Ewelina Sadowska.
- 12:10–12:25: Specialized large language model agents for fine-grained evidence extraction in evidence synthesis. The study evaluates task-specific agents for patient-flow extraction from clinical studies and for identifying suspected drugs in pharmacovigilance reports, with each value linked to its source. Authors: Joanna Konieczna, Ewelina Sadowska, Monika Opalek, Artur Nowak, Ewa Borowiack and Damian Stachura.
Both studies address the same practical question from different directions: how the way a task is specified for a model affects whether reviewers can check and rely on its output.
Workshop: from summary of findings to recommendations
On Wednesday 9 December, 11:00–12:30 (ICE Kraków, S4 C), Cochrane, Evidence Prime and the MAGIC Evidence Ecosystem Foundation run a joint workshop, Streamline evidence-to-recommendation workflows using structured summary of findings. Participants will update a Cochrane summary of findings table in RevMan and carry it into GRADEpro and MAGICapp to start a guideline. The workshop was developed by Rebecka Hall (Cochrane), Bartłomiej Dietl (Evidence Prime) and Linn Brandt (MAGIC).
The aim is practical: evidence prepared in a review should not have to be rebuilt at each step on its way to a recommendation.
Chairing Responsible AI sessions
Artur Nowak will also co-chair four oral sessions in the Responsible AI in evidence synthesis stream, all in ICE Kraków, S2 Theatre:
- Tuesday 8 December, 16:00: Responsible AI in evidence synthesis 3
- Wednesday 9 December, 11:00: Responsible AI in evidence synthesis 5
- Wednesday 9 December, 14:00: Responsible AI in evidence synthesis 8
- Thursday 10 December, 11:00: Responsible AI in evidence synthesis 12
Session times, rooms and speakers are listed in the official Colloquium programme, which remains the authoritative source if anything changes.
Meet us in Kraków
If you are attending and would like to talk about AI-assisted evidence synthesis, living evidence or guideline development in GRADEpro, please get in touch. We are happy to arrange a meeting during the Colloquium.
