Your slide.
Your call.
A guided workspace for soil-microscopy assessment that keeps the judgement where it belongs — with you — and makes the report you hand a client defensible enough to survive being questioned.
- Present
Bacteria
reviewed · 12 FOV
- Present
Fungi
reviewed · 12 FOV
- Present
Protozoa — flagellates
reviewed · 12 FOV
- Absent
Protozoa — ciliates
reviewed · 12 FOV
- Not assessable
Nematodes
sample degraded
- Undecided
Oomycetes
4 of 12 FOV
Illustrative class review — each organism class worked through to a decision. The report stays blocked while any one of them is undecided.
Five steps, one screen at a time.
- 01
Sample
Register the sample and its context once. No spreadsheet running alongside.
- 02
Upload
Fields of view, in the order you shot them.
- 03
Annotate
One organism class at a time, each worked to a decision — present, absent, or not assessable, with a reason.
- 04
Counts
Bacterial and nematode counts recorded against the same session.
- 05
Report
A client-ready report, generated only once the assessment is genuinely complete.
Beneficials and pathogens alike.
The organism classes an assessment works through, one at a time. Real microscopy from open datasets, with the original attribution preserved beside each plate.
Sample 78Trichoderma harzianum
Beneficial inoculant · biocontrol agent
Zenodo 10.5281/zenodo.7965200 · CC-BY 4.0
Sample 04Fusarium spp.
Soilborne pathogen · biocontrol target
Zenodo 10.5281/zenodo.7965200 · CC-BY 4.0
Sample 106Verticillium dahliae
Soilborne pathogen · strawberry / cotton
Zenodo 10.5281/zenodo.7965200 · CC-BY 4.0
Sample 64Phytophthora spp.
Oomycete pathogen · potato / forestry
Zenodo 10.5281/zenodo.7965200 · CC-BY 4.0
A report that cannot quietly be wrong.
Complete, or it does not leave
A report will not generate while any required organism class on any included field of view is still undecided. Half-reviewed slides do not become client documents.
Current, provably
A report is released only when a fresh check agrees it reflects your latest edit. Change a count after generating one and the old version stops being downloadable, rather than silently going stale.
Yours alone
Two tabs or two devices cannot overwrite each other's decisions. The work you signed off is the work that ships.
The software does the bookkeeping. You do the biology.
Assisted detection proposes; you dispose. Every proposal is something you accept, correct, or reject — and your correction is the record, not a suggestion the system is free to overrule.
We are equally explicit about what the assist is not. It is not a verdict, and we do not publish an accuracy claim we have not earned. What is proven today is narrower and more useful: the workflow cannot lose your work, cannot ship an incomplete assessment, and cannot hand a client a stale one.
What it does not do.
It does not sign your name
The assessment is yours. We do not issue an opinion on your behalf, and the report is clear about who made the call.
It does not quantify what we cannot
Some measures are simply not offered, because we cannot stand behind them yet. Absent is better than approximate.
It does not replace your microscope
It is the workspace around it — capture, decide, record, report — so the expertise you already have leaves the room in a form a client can act on.
Put a real slide through it.
Early access is open to practising consultants and labs. The most useful first session is a sample you have already assessed yourself — so you can see whether the record matches your judgement.