Problem

Hospitals are deploying clinical AI from many vendors, particularly in medical imaging, faster than they can oversee it. Once a model is live, hospitals often cannot tell whether it is still working: scanner upgrades, changing patient populations, and other shifts can cause model drift, while accuracy may vary across patient groups. The result is a patient-safety, compliance, and operational risk that falls primarily on the hospital rather than the AI vendor. The central use case is giving radiology and hospital leadership visibility into whether deployed imaging models still agree with radiologists and perform reliably across populations.

The pain is less about buying another predictive model than about proving that existing models remain safe and effective after deployment. Lattice addresses the oversight gap created when models operate inside PACS and EHR environments without a shared view of accuracy, drift, fairness, latency, or silent vendor updates. The available materials do not quantify a specific dollar cost, but they make clear that the economic exposure is the cost of unchecked clinical risk, regulatory scrutiny, and potentially degraded care.

Product / Service

Lattice positions itself as AI governance delivered as a managed service for hospitals: a monitoring and governance layer that sits beside the existing clinical stack. It ingests HL7 and DICOM signals from deployed models without requiring a rip-and-replace, continuously measures population drift, subgroup disparity, latency, and silent updates, and incorporates aggregate radiologist thumbs-up or thumbs-down feedback. It then turns those signals into evidence for safety, governance, and compliance.

The delivery model is deliberately operational rather than dashboard-centric. Lattice sends signed, role-specific PDF reports to the appropriate inbox each morning, with alerts available through email, Microsoft Teams, and SMS, and can produce chain-of-custody evidence packets when compliance or counsel asks. It is designed for on-premises or private-cloud deployment, with PHI remaining behind the hospital firewall and no PHI required on the wire. The benefit is a verifiable daily operating record of what clinical AI is doing, without changing how clinicians work or placing Lattice in the care pathway.

Market

Lattice competes in healthcare AI governance, clinical AI observability, and post-deployment monitoring, with a particularly sharp focus on deployed medical-imaging AI. Its differentiation is the combination of continuous drift and fairness monitoring, radiologist feedback, signed evidence, and a managed reporting workflow for hospital leadership rather than a generic model dashboard. Adjacent competitors and alternatives include Solytics Partners, whose healthcare governance platform covers inventory, validation, monitoring, observability, and compliance; Censinet RiskOps, which tracks AI-related risk and routes findings to governance stakeholders; and broader platforms such as ModelOp and Signal1 that provide model inventory, lifecycle governance, and performance-drift monitoring.

Lattice appears to be at an early commercial or pilot stage. Y Combinator lists it as an active Spring 2026 healthcare company in San Francisco with a team size of one, while the company website prominently invites hospitals to request a pilot. No named customers, revenue figures, or funding details are disclosed in the reviewed materials, so it is best characterized as pre-scale and likely pre-revenue or pilot-revenue rather than an established vendor with publicly demonstrated traction.