Clinical review work for your spare minutes
Konsil is a planned service that pays verified doctors to review synthetic and anonymised cases on their phone, one question at a time.
Building medical AI? See what Konsil offers companies
One task looks like this. Answer it as you would on the train.
Ist die Medikationsliste vollständig?
Is the medication list complete?
Synthetic case, no patient data
Entlassbrief
Patientin, 78 Jahre. Hospitalisation 12. bis 18. September.
Diagnosen
- Kardiale Dekompensation bei hypertensiver Herzkrankheit
- Neu diagnostiziertes Vorhofflimmern
- Arterielle Hypertonie
- Diabetes mellitus Typ 2
Verlauf
Rekompensation unter intravenöser Diuretikatherapie mit Gewichtsabnahme von 4 kg. Bei neu diagnostiziertem Vorhofflimmern Beginn einer oralen Antikoagulation mit Apixaban 5 mg zweimal täglich sowie Frequenzkontrolle mit Bisoprolol. Austritt in gebessertem Allgemeinzustand nach Hause.
Medikation bei Austritt
- Torasemid 10 mg1-0-0
- Bisoprolol 5 mg1-0-0
- Lisinopril 10 mg1-0-0
- Metformin 1000 mg1-0-1
Procedere
Tägliche Gewichtskontrolle. Kontrolle von Kreatinin und Kalium beim Hausarzt in einer Woche.
“Unsicher” is a valid answer.
How it would work for you
Sign up and verify once
You confirm your credentials, your specialty and your languages.
Open a task when you have a few minutes
Each task is one case and one focused question, in German, French or Italian.
Several doctors answer independently
A statistical model combines the answers and asks for another opinion only while the case is still uncertain. Hidden cases with known answers measure how reliable each reviewer is.
Get paid per task
A fixed fee per task, with a bonus for accuracy. You see your earnings and your own accuracy right away. Fees are not set yet: we are asking doctors what would make it worth their time.
What Konsil deliberately does not do
No identifiable patient data
Only anonymised or synthetic cases. The pilot uses synthetic cases only.
Nothing stored on your phone
Cases will be view-only, and hosting in Switzerland is planned. You never see cases from your own hospital.
No patient care
Your answers train and test AI systems. They do not decide any patient’s treatment.
No shifts
You choose when and how much, outside working hours.
We are six students at the University of Bern. Nothing here is built yet. Before we build it, we want your honest view.
Konsil is our project for the Introduction to AI course in autumn 2026. We would like to know when you would realistically have time for this, what would make it worth doing, and what would stop you from trying it.
Questions, or something we got wrong? Write to Felix Kunz, felix.kunz@students.unibe.ch.
Join the waitlist
We will write to you when the pilot with synthetic German clinical texts starts. If you tick the box, we will also ask you for a short conversation.
For teams building medical AI
One label per case, with a confidence score and a record of how it was reached.
What you would receive for a case like the one above
Example record. Synthetic case, simulated reviewers.
- Case
- Discharge letter, German, synthetic
- Question
- Ist die Medikationsliste vollständig?
- Label
- Nein, the list is incomplete
- Confidence
- 97%
- Opinions
- Nein, Ja, Nein, Nein
- Closed
- after 4 of 5 possible opinions, target 95%
- Practising clinicians, in the local language
- German, French and Italian clinical material, reviewed by verified doctors, residents and advanced medical students.
- A confidence for every label
- Several clinicians answer each case independently. You see how sure the model is and how many opinions it took.
- Opinions only where they are needed
- Our proof of concept has to show that the model reaches the accuracy of fixed majority voting with fewer paid opinions. We have not shown it yet.
- Made for documentation duties
- Planned: hosting in Switzerland and a record of how each label was reached, for teams that document data governance under the EU AI Act.
Ask for a pilot batch
We are looking for one team to run a first batch of 200 synthetic German clinical texts with us.