One case, four workflows, one viewer.

A pathologist used the same microscope for a clinical case, a consult, a teaching session and a research question. Twenty years into digital pathology, each of those jobs tends to live in its own system, with its own login and its own copy of the slide. Scanning turned out to be the easy part; connecting the work is the hard part.

The worklist arrives already screened by a pan-cancer AI algorithm, so the one suspicious case is read first. In the viewer the pathologist opens the case information, the radiology, gross and clinical photographs and the prior biopsy, and lays up to four slides side by side. The fellow's annotated mitoses and questions sit on the slide itself. Stains are ordered from the side panel, a notification arrives when they are ready, and the tumor area is marked for molecular testing.

Without a portal, glass slides are packed, shipped, accessioned, read and shipped back, with a risk of breakage or loss along the way. Here the case arrives in the consultant's own worklist through their LIS, with the referring pathologist's markings, gross photographs, stains and report already attached. The two pathologists chat directly, the consultant requests a DUX4 stain and FISH, and the results come back into the same viewer. The signed consult report flows back to the referring institution.

A region of the slide can be searched against similar-looking cases, the way an image search works. Beyond one institution, an AI data commons pools de-identified data under one governed, searchable index while every contributor keeps ownership in its own tenant. The demo used mock data from 35 institutions: a query for patients under 18 with this sarcoma, an access request for mitotic rate and survival, a supervised study with fixed regions of interest and labels for each mitotic figure type, annotation and review, export, and a model trained to score the index case.

With the free education model, whole slide images sit inside the lecture itself, annotated so a beginner knows what to look at in what otherwise looks like a pink ocean of blue dots. Two slides can be compared on the same page, a case is shared with a class in about 30 seconds, and nobody has to keep a hundred sets of fading glass. The same link lets a pathologist sit with a patient and show them the images their diagnosis came from.

The workshop closed with what is coming: an intelligence layer in the back end, a framework for building agents that track cases through the system and summarize them for the pathologist. It was announced, not shown, so on this page it is drawn dashed. The closing point was the one that started the session: pathology has always been done together, at the next microscope or the multi-headed scope, and the digital infrastructure should keep it that way.