Questions people ask us

Researchers in the pre-publication phase — PhD candidates and postdocs first: the people at the bench who hit a failed protocol and need credible context fast. Research leaders, R&D directors, universities and small biotech companies are the next circle.

Experimental, data-heavy sciences first — life and health sciences in particular — where methods, conditions and failed runs matter most. Natural sciences, engineering and agricultural research follow the same pattern and are in scope.

We start with Europe and North America, home to roughly 3.5 million full-time researchers. Anyone can read and ask; our first communities and pilots are there.

Not individual researchers: they use SciCollab for free. Institutions and R&D organisations pay for private, governed workspaces for their labs — with audit logs, single sign-on and support for open-science compliance. Pricing is not final yet; we are setting it with our first pilot institutions.

When an experiment deviates from the protocol, today’s path is a chain of dead ends: ask a peer (nobody knows), ask the supervisor (stumped or busy), e-mail collaborators (no answer) — and the work disappears. SciCollab shortens the time to a first useful answer and keeps what was learned, including what did not work.

Every researcher in our second survey asks AI first when an experiment fails — and their success rates ranged from 5% to 90%. AI has no access to the “dirty” context: the exact conditions, the failed runs, the unpublished data. SciCollab uses AI to find and structure prior cases; people with hands-on experience validate them.

Because a failure that nobody records gets repeated. In our survey, 71% of researchers said they would share negative results, yet none of them follows an existing negative-results platform — a journal is the wrong place for them. On SciCollab, a null result is part of the normal workflow: citable, searchable, and credited on your profile.

Each solves one piece:

• ResearchGate — networking and papers, but no data sharing and no troubleshooting.
• Stack Overflow — fast answers, but built for code, not for the lab.
• GitHub — excellent versioning, but you need to code, and it gives no academic credit.
• Electronic lab notebooks — good data management, but the knowledge stays inside one lab.

A structured, versioned record of how something was actually done — materials, conditions, outcome — including the conditions under which it failed. Others can reuse (“fork”) it, and their results attach to it, whether they worked or not.

Reputation that reflects real contribution — answers, methods, reviews, negative results — kept as separate measures rather than one score. Credit for co-authorship is agreed with the project owner; see the Terms.

Yes, for individual researchers. Paid plans are for institutions that want private workspaces for their teams.

We are preparing an invite-only beta with a first cohort of about 50 researchers, before a public launch targeted for December 2026. To be on the list, write to us.

Yes. A pilot usually starts with one to three labs and a local champion, then widens. Tell us about your lab.

What is in your own workspace stays yours. What you add to someone else’s thread or project is free to reuse (CC0) unless you agree credit with them first. Details in the Terms & Conditions (draft).

Not in a public area. Under the European Patent Convention, publishing an invention can destroy its novelty — there is no general grace period. Keep that work in a private workspace. See clause 2.1.

Our target is cloud infrastructure in the European Union, for data sovereignty and GDPR compliance. The preview you can see today runs on example data only — no real research data is stored in it.