Gestalt Diagnostics

An image management and workflow platform, with its own learned tools beside the partners' models. PathFlow centralizes slides, cases and patient data, routes cases, integrates with laboratory systems and supports sign-out, education and research; none of that needs a model. Most diagnostic algorithms on it come from partners the company names, including Deep Bio, Ibex, Lunit, Mindpeak, Paige, Primaa and Visiopharm, and those are the partners' AI, not Gestalt's. Gestalt's own learned components are AIRE, a requisition engine that reads forms including handwriting and learns each new form format, algorithms it describes as Gestalt-developed in its AI portfolio, and custom models it builds with customers.

Artificial intelligence is a feature layer on a product whose value stands without it.

A customer-set threshold governs the one automated action, with the consequence of an error unaddressed. AIRE accessions cases automatically only above a confidence threshold the laboratory sets, up to 95 percent, and flags information below it for a person to correct. The AI Algorithm Evaluator lets pathologists score and compare algorithms side by side on their own cases, track scores over time and keep an audit trail of why an algorithm was chosen. The company describes its AI as keeping pathologists in control of diagnosis. Nothing says what happens when AIRE accessions a specimen to the wrong patient above threshold, how such errors are detected, or what the platform requires of partner algorithms before offering them.

The oversight structure described, with the handling of an error missing.

The company's own models are described by function only. AIRE is described as a continual learning platform that trains on each new requisition format and interprets handwriting and misoriented insurance cards; the AI portfolio mixes Gestalt-developed and third-party algorithms without marking which is which; and custom models are co-developed with customers. No model type, training data, version or update practice is published for any Gestalt model, and nothing says how AIRE's continual learning is controlled or validated before it changes accessioning behavior. The company lists Anthropic among its technology partners without describing any use.

Architecture described in general terms with nothing identified.

The model providers for hosted algorithms are named, with the data terms between them unstated. The partners page names the AI companies whose algorithms run on PathFlow, including AIRA Matrix, Deep Bio, Flagship Biosciences, Ibex, Lunit, Mindpeak, Paige, Primaa, Visiopharm and VistaPath, alongside scanner, laboratory system and record system partners, and lists Anthropic and NVIDIA among technology partners. Gestalt's own models, including AIRE, are built in house. What is missing is the chain for patient data: whether partner algorithms run inside the customer's PathFlow environment or receive images on their own infrastructure, what partners may retain, and which cloud host serves cloud deployments.

Substantial partial disclosure, with the model providers named.

Named customers and ceiling figures, with no study. Customers named with attributed quotes include Tulsa Medical Laboratory, Pathology Diagnostics Services Group, BioReference Laboratories and the Indiana Animal Disease Diagnostic Laboratory at Purdue. AIRE's figures are stated as ceilings, accessioning time reduced by up to 80 percent and automatic accessioning at a customer-set threshold of up to 95 percent accuracy, with no sample or method. The company reports first place in two global AI competitions, which shows technical capability on a benchmark task, not performance in a laboratory. The company cites no peer reviewed or independent evaluation of PathFlow, AIRE or Gestalt's algorithms.

None of the figures comes with a method or a denominator.

General assurances, and a model that learns from customer documents. The company says its platform safeguards protected health information with end-to-end encryption and is built to HIPAA standards. It also describes AIRE as a continual learning engine that trains on each new requisition format, and requisitions carry patient identifiers and insurance details. Nothing says whether that learning stays within one laboratory's deployment or is pooled across customers, what is retained, or whether a laboratory can opt out, and nothing describes what partner algorithms on the platform may retain from patient images. The privacy policy covers the website and online resources. No safety engineering is published.

General assurances that do not answer what reaches a model.

HIPAA compliance is claimed without the document behind it. The company describes its platform as built to HIPAA, SOC 2 and CE-IVD standards and as safeguarding protected health information, and offers cloud deployment, image sharing and outsourced laboratory system support, all of which can involve patient data. The only privacy policy covers the website and online resources, and no business associate agreement or statement of when Gestalt acts as a business associate is published.

Compliance claimed without the underlying document.

A recognized attestation named without the report. The company states that its platform is HIPAA, SOC 2 and ISO certified with independent validation of security and privacy controls, and separately refers to ISO-certified infrastructure, which may be its host's. No SOC 2 report type or period, ISO standard number, certificate or trust center is published, so which certification belongs to Gestalt and which to its hosting provider cannot be told apart.

A recognized certification named, with scope to raise.

The platform carries a European mark and the AI's position is stated. The company states that the PathFlow Viewer and Pathologist Workflow application are CE certified under the IVD directive, which covers the platform rather than any algorithm.

Its AI portfolio and diagnostic claims carry the footnote that they are for research use only in the United States unless a laboratory follows College of American Pathologists guidance for laboratory developed tests, which places validation on the laboratory. One partner application is described as CE-IVD certified, which is the partner's position. FDA's databases list no Gestalt clearance.

Responsible use is committed to, and the evaluation tool is the customer's to run. The company says it helps laboratories use AI responsibly and evolves with emerging AI standards, and its AI Algorithm Evaluator lets a laboratory compare algorithms on its own cases with customizable scoring, historical score tracking and an audit trail, which is how a laboratory can test a model on its own population. Nothing is published about Gestalt's own models: no evaluation of AIRE across referring practices, form types or handwriting quality, where misreads would slow some practices' cases more than others, and no results for Gestalt-developed image algorithms.

Policy commitment with no evaluation of the company's own models.

Review and audit mechanisms exist, and nothing stands behind an automated outcome. AIRE flags information below the laboratory's confidence threshold for correction before accessioning, and the Algorithm Evaluator keeps an audit trail of how algorithms were scored and chosen. Above threshold, AIRE creates or matches patient records unattended, and an accession to the wrong patient carries through to the diagnosis and report; no error rate is published for a given threshold. For partner algorithms hosted on PathFlow, nothing states how responsibility is shared between the partner, Gestalt and the laboratory. There is no warranty, indemnity or remediation commitment.

Major record systems are named, without integration documentation or proof of a two-way connection. The company states integrations with Epic, Cerner and major and proprietary laboratory and record systems, lists Epic and Oracle Health among its partners, and describes automated case routing with laboratory system integration, a Health Data Cloud interface engine connecting laboratory systems, records and reporting agencies, and electronic laboratory reporting to public health agencies. A customer quote describes consolidating cases from clients on different laboratory systems into one workflow. No marketplace listing, interface specification or statement of which data flow in which direction is published.

Named systems, with direction and production evidence unpublished.

Cloud or on premises deployment, with the cloud route's location unstated. PathFlow is offered as a cloud-based or on-premises deployment without architectural change between single-site and enterprise installations, and PathCloud provides web-based image sharing. On premises keeps images at the laboratory. For the cloud route no host, region, tenant isolation or retention is stated, and the company's IT services and outsourced laboratory system support add routes by which its staff can reach customer systems.

Options stated with residency committed only for the on premises route.

No price is published, and the product structure is legible. PathFlow is modular, with clinical, education and research modules adopted separately, PathCloud as a lighter web-based option, an AI portfolio of Gestalt and partner algorithms, custom AI development, and IT services including laboratory system support, migrations, electronic laboratory reporting and staff augmentation. AIRE is offered as a fully managed system or a standalone module. No license basis, per case or per slide rate, or partner algorithm pricing is published.

The sales posture is discoverable, with no price.

Clinical, education and research pathology are named, with deployments behind them. PathFlow supports clinical anatomic pathology workflows, an education module for training, proficiency testing and credentialing, and a research module, for health systems, academic centers and private laboratories, and it is used in veterinary diagnostics at Purdue's Indiana Animal Disease Diagnostic Laboratory. It acquires images from many scanner makers the company names, including Hamamatsu, Leica, Philips, 3DHistech and Grundium. The viewer's European mark supports clinical use there; the AI is research use in the United States unless validated by the laboratory.

Coverage named with validation behind part of it.