app.causalmap.app
Turn interviews, reports and open survey answers into maps of what people say causes what. Code causal claims by hand or with AI, then filter, compare groups and summarise.
European Evaluation Society 2026 · Lille
We are presenting two sessions at the 16th EES Biennial Conference, “Evaluation for Vibrant Democracies”, 26 to 30 October 2026.
Here are the tools, and below is information on the sessions.
qualiainterviews
A conversational AI interviewer for exploratory qualitative interviews at scale. Run interviews, then bring structured outputs into causal mapping when you are ready.
Rubicon
A free, open-source plugin for Claude with which evaluators answer their questions from a folder of documents. You agree the analysis before anything is read; fixed rules count every number and check every quotation.
Our sessions
Both sessions are on Wednesday 28 October.
-
Interrogating evaluation: a live experiment and conversation, with a little help from AI: Innovative Modality · Session S1d, Strand S01 New Methods · 3.7 The Studio · 13:00-14:30 · Silva Ferretti (Independent Consultant), Steve Powell (Causal Map Ltd)
Participants are interviewed live by Qualia, an AI interviewer, about where evaluation is going, what new capacities matter most, and what risks evaluators should now take seriously. Qualia then synthesises the interviews. The session treats that synthesis as a starting point: small groups discuss the same questions, some with AI support and some without, and in plenary we compare what each way of working made easier, harder, more visible or less visible.
-
Non-mosquitoes do not cause non-mosquito bites: Individual paper · Session S3e "From Stories to Data", Strand S03 Responsiveness · 3.1 C · 16:00, in the session running 15:00-16:30 · Steve Powell (Causal Map Ltd), Fiona Remnant (Bath SDR / QuIP)
People mostly do not speak in variables, yet most ways of coding "what causes what" assume they do. We argue that when you are modelling what people believe, the default should be to code causal claims as the speaker made them: mosquitoes cause mosquito bites, with opposite labels kept separate and combined only when the analysis needs it. This keeps the stakeholder's logic rather than the analyst's, and it is also the easiest coding for people and for AI. The full conventions are in Minimalist coding for causal mapping and Combining opposites.
An informal chat on Thursday
Thursday 29 October, from about 16:30, at the café tables near Room 0.2, straight after session S13e. About twenty minutes; anyone is welcome, just turn up.
It is for anyone interested in getting Claude, or their favourite model, to follow an explicit evaluation method, in steps that somebody else could check or rerun. We will bring a couple of ideas of our own (an open format for recording an analysis, and Rubicon) and want to hear what other people are already doing. Background: Pilots, not passengers.
Get in touch
If we met in Lille and you would like to follow up, write to hello@causalmap.app.