SciCollab is where researchers share the data, methods and failed runs behind their work — before publication — and get help from peers who have been there.
Why we exist
An experiment fails. The researcher asks a colleague — nobody knows. The supervisor — stumped, or busy. Collaborators by e-mail — no answer. Weeks later, the work is shelved, and the next lab repeats the same failure, because nobody wrote it down anywhere they could find it.
70%+
of researchers surveyed by Nature have tried and failed to reproduce another scientist's experiment.
Baker, Nature 533, 2016
71%
of the researchers in our own survey would share their negative results — yet none of them follows an existing negative-results platform.
SciCollab survey, n = 17
5–90%
success rate of AI tools on failed experiments, for researchers who all ask AI first. It has no access to the messy, unpublished context.
SciCollab survey, n = 6
What we are building
01Questions that carry their evidence
A question comes with the data, the code, the conditions and the outcome that produced it — so the people answering see what you see.
02Method cards
Structured, versioned records of how something was actually done, including the conditions under which it failed. Fork one, run it, and your result attaches to it.
03Negative results as first-class objects
A failure that reproduces is a contribution. Null results are citable and reusable, with the same weight as a positive one.
04A living database of unpublished science
Raw data and methods stored before and alongside formal publication — searchable by the next person who hits the same wall.
05AI that finds; people who validate
AI retrieves and structures prior cases. The community checks them. No black-box answers on questions that matter.
Where we are
We interviewed and surveyed researchers across disciplines, and changed course on what we heard: the real pain is not publishing — it is being alone at the bench with a problem. A first prototype was tested with early users; their verdict was that the value appears when data, methods and discussion live in one place.
DONE
Field research
Interviews and two surveys with researchers.
DONE
Prototype
Five core flows tested with early users.
NEXT
Invite-only beta
A first cohort of about 50 researchers.
PLANNED
Public launch
Targeted for December 2026.
Who we are
Six people who met in the ESSEC & Mannheim Executive MBA, where SciCollab began as our strategic project. Between us: a clinician-scientist, engineers, and people from operations, finance, marketing and enterprise sales.
CChristian
Product & Research Lead
Clinician-scientist. Domain expertise, R&D roadmap, academic partnerships.
JJanmejay
Tech & Engineering Lead
Platform architecture, the MVP build, structuring data, the AI retrieval layer.
EElias
Marketing & Community Lead
Go-to-market, community growth, the lab-champion network, conference adoption.
YYann
Operations & Legal Lead
Infrastructure, compliance, GDPR, data governance, the IP framework.
HHoelig
Finance & Strategy Lead
Financial modelling, investor relations, pricing strategy, SaaS metrics.
MMaureen
Partnerships & Enterprise Sales
Institutional pilots, letters of intent, enterprise onboarding, R&D partnerships.
Survey figures are from our own field research in 2026 (small samples, as noted). Nature figure: M. Baker, “1,500 scientists lift the lid on reproducibility”, Nature 533, 452–454 (2016).