Jonas reviews research while it is happening, asks when the evidence or reasoning behind a decision is missing, and keeps a traceable record of how the work was done. PIs spend their time on the decisions that actually need their judgment.
Pilot Jonas with your teamA junior researcher can now generate a literature review, an analysis, and a draft in a day. The volume of work went up. The capacity to verify it did not.
So the PI becomes the bottleneck. Every claim, source, and analysis decision still has to be checked by the one person who cannot delegate that check, and the reasoning behind most of it was never written down in the first place.
What the PI still has to check, by hand
Claims
Is the conclusion actually supported by what was run?
Citations
Does the cited paper say what the draft says it says?
Methodology
Which thresholds, definitions, and windows were used?
Analysis choices
What was cleaned, excluded, or reran, and why?
Reasoning
Was the criterion set before or after seeing the result?
Lab notebooks and documentation tools give researchers somewhere to put this. They still depend on the researcher knowing what will matter later and remembering to write it down.
Jonas notices what the researcher forgot to mention, and asks.
Documentation stops being a separate task. It becomes the record of the questions a researcher had to answer while doing the work.
Researchers keep doing analyses, experiments, literature work, and writing in the tools they already use. There is no new workspace to move into.
It reviews what changed and what supports it, then asks the researcher while the context still exists. If the work is well documented, it stays quiet.
Routine checks resolve between Jonas and the researcher. Only decisions needing senior scientific judgment get escalated.
Analysis review
What changed
Cohort dropped from 84 to 73. Association no longer significant.
What Jonas checked
9 samples
Below the QC threshold defined in the protocol
2 samples
Duplicate sample IDs, documented in the cleaning log
Open
Were the criteria set before the result was seen?
Researcher message
Jonas · 4:52 PM
You removed 11 samples before rerunning. Which exclusion criteria did you use?
Researcher · 5:04 PM
9 failed the QC thresholds in the protocol. 2 had duplicate sample IDs.
Saved to the analysis record
QC exclusions linked to the protocol section that defines them. Duplicate IDs linked to the cleaning log. Whether the criteria were prespecified goes to the PI.
PI view
Analysis reviewed. 1 methodological decision needs your judgment.
Documentation software stores what researchers remember to tell it. Jonas asks about the reasoning that would otherwise stay in someone’s head.
Jonas asks about consequential decisions at the moment they are made, not months later when nobody remembers why.
Jonas · 4:52 PM
The effect size changed after the rerun. What differs from the previous version of this analysis?
You · 5:04 PM
Switched to the per-protocol population and dropped the imputed values.
Claims and decisions stay linked to the material behind them. When a source does not support the claim, Jonas says so.
“Elevated IL-6 predicts progression in this population.”
Cited source · results, para 4
“…IL-6 was associated with progression in the subgroup with baseline fibrosis…”
Routine checks resolve between Jonas and the researcher. The PI sees the decisions that actually need senior judgment.
Resolved with the researcher
Missing values checked and documented
Resolved with the researcher
Duplicate IDs removed, criteria linked
Needs your judgment
Exclusion criteria may be post hoc
The trail comes from questions researchers had to answer anyway. Nobody writes documentation twice, and handoffs explain the why, not just the what.
Cohort defined
Aug 4Protocol §2, eligibility list attached
11 samples excluded
Aug 6QC thresholds + duplicate IDs, approved by PI
Model rerun
Aug 7Association not significant, prior version kept
Retrospective chart review is the same shape as the rest of Jonas: read the protocol, pull the required variables from records, link every value to the note it came from, and ask a researcher when a case needs judgment. The output carries its own provenance.
No new platform to adopt, no project board to keep updated. Jonas reads the work your team is already producing and replies in the same place.
Teams keep their analyses, notes, and drafts where they already are. Pilots today run on a shared document the team writes into as work progresses.
Questions land next to the work as comments, or by email when a researcher is away from the document. Only when something is missing, unsupported, or ambiguous.
The research record: what was done, what supported it, which decisions were made, who approved them, and what is still open. Yours to export and keep.
Early pilots are run hands on. We read the team’s work and ask the questions ourselves while we learn which ones actually change the science. Everything else gets automated only after it earns its place.
It reviews research work as it happens: analyses, experiments, literature work, drafts. When the evidence, criteria, or reasoning behind something is missing or does not hold up, Jonas asks the researcher and keeps the answer with the work.
No. A notebook stores what a researcher remembers to enter, which is exactly the part that fails. Jonas works the other direction: it reads what the team produced and asks about what is missing. The record is a byproduct.
Only on exceptions. If the work is documented and supported, Jonas stays quiet. Questions go to the researcher who made the decision, and anything needing senior scientific judgment goes to the PI.
No. It separates evidence from suggestion and never resolves a judgment call on its own. Researchers decide, Jonas records the decision and what supported it, and the PI approves what matters.
Deliberately lightweight. Your team writes updates about what they did, what they found, and what they decided in a shared document. We review those entries and ask the missing questions, by hand at first, so we learn which questions change the science before automating them.
No. The problem is general to scientific teams. Our first users are biomedical because that is where our network is, and chart review is one workflow Jonas supports, but the core of it is analysis, evidence, and reasoning in any research team.
Pilots can start on non-patient scientific work, which keeps the setup simple. For chart-review work we use de-identified or institution-approved records only, a researcher reviews every value, and the trail records what changed, who approved it, and the evidence behind it. We do not claim certifications we have not completed.
No. Today Jonas checks the work. Over time it takes on bounded research tasks under the same rules: evidence attached, humans accountable for judgment. The point is a PI supervising more research, not fewer people doing it.
Pilot teams use Jonas at no cost while we build the product around their work. Paid plans come later, and pilot teams get them first.
Run Jonas alongside your team for a few weeks. Your researchers keep working the way they work, and you keep the research record it builds.
Pick a time below. Twenty minutes, and bring something the team is working on now.