Status and history
of incidents
The operational state of the platform's services and the public record of incidents and maintenance. Contracting institutions also receive direct communication by email.
Measured services
Each row is a real request to the public address, with the response code and the time until the full page loads. Since the snapshot is more than 36 hours old, it is historical and does not represent the current operational state.
- SITE AND PUBLIC PAGES operational in the snapshot · HTTP 200 · 744 ms The site's pages, including this one. www.laudos.ai
- REPORTING PLATFORM operational in the snapshot · HTTP 200 · 1005 ms The application where the radiologist dictates, reviews and signs. copilot.laudos.ai
- RADCOMMONS operational in the snapshot · HTTP 200 · 705 ms The classification corpus and the API that GUIDE queries. radcommons.laudos.ai
- LAIBENCH operational in the snapshot · HTTP 200 · 764 ms The public benchmark, case by case. laibench.laudos.ai
- TECHNICAL DOCUMENTATION operational in the snapshot · HTTP 200 · 465 ms The integration manual and the API reference. manual.laudos.ai
How this table is built. An HTTP GET request to each public address, best of 3 spaced attempts, 15 s limit. Run on every site publish and once a day by GitHub Actions. The result of each check is versioned in the site repository (content/status.json), with the check stored for the last 90 days. We do not infer uptime between snapshots.
Dictation, RIS/PACS integrations and the manager dashboard do not appear here. All three require credentials, and no one can check them from outside. Laudos.AI communicates their status by email to contracting institutions. Any degradation goes into the history below.
Incident and maintenance history
Instability in report generation
Report generation was unstable for about 1 hour in the late afternoon. The failure occurred at the moment of generation: the report was not created and the interface returned an error. A new attempt usually completed successfully.
During that period, some attempts returned an error before the report was created. The failure did not reach any report already saved and no data was lost. We have already returned to your plan any usage counted on an attempt that returned an error, with no impact on the limit.
We fixed the cause at its source and the platform is back to normal. We now retry automatically, so that momentary instability does not reach the user.
User reports were what pointed the way to the cause. Anyone who hits an error when generating can write to oi@laudos.ai.
Occasionally malformed text output in STT
Monitoring identified an occasionally malformed text output in the speech transcription step. The team applied the fix and checked the workflow. This public record does not include report content, dictated terms, user identification or patient data.
Database update and upgrade
Scheduled maintenance window for the update and upgrade of the platform database, communicated in advance to institutions. During the window, the platform was offline: dictation, signature and report delivery unavailable. The window did not affect any report, template or audit trail.
No other incidents recorded in the period. Partial degradations also count as incidents and go into this history.
How we communicate incidents
Institutions with an active contract receive direct communication by email in case of an incident or scheduled maintenance, with scope, impact and expected time to normal. After relevant incidents, a root cause report is sent to those affected.
Laudos.AI schedules maintenance with potential impact in a window agreed with institutions, and announces each window in advance.
The life cycle of an incident
The cycle has five stages: detection, containment, proactive communication to those affected, normalization and a root cause report with preventive actions. Detection comes from continuous monitoring of availability, dictation latency and integration queues. The same per-exam log that supports clinical auditing supports the technical reconstruction of the incident.
Partial degradations count: if dictation is slow or report delivery is delayed, it is an incident, even with the platform up.
Availability targets
The institutional contract defines the availability targets and associated credits, service by service. The public criterion on this page reflects the current operational state; the contractual criterion reflects the commitment.
Direct questions
What counts as an incident?
Any degradation that affects reporting: unavailability, abnormally slow dictation, integration failure or delay in delivering the signed report.
What is the fastest channel for reporting a problem?
The email oi@laudos.ai with the subject "Incident". Institutions in production also have a direct channel agreed during deployment.
Read the current state in the right context
What this page can state
The current state reports the known condition of the published services and the incidents recorded here. It does not replace the contractual SLA. No one turns this page into a historical percentage without a published series. Nor does it prove the behavior of a specific integration at the institution.
When an incident occurs, the update needs to separate impact, affected services, start, mitigation and resolution. Institutions with an integrated workflow also use the operational channel in the contract. A local RIS, PACS or credential failure may not appear as a general platform incident.
Platform available does not mean the full workflow is available
A report passes through the browser, authentication, processing, integration and the institution's systems. The status page covers only the services published by Laudos.AI. PACS, RIS, the local network and external vendors fall outside that perimeter. In an incident, the investigation needs to locate at which stretch the exam stopped before assigning a cause.
Uptime history, latency or recovery commitment should only be used when published for the corresponding period and service or defined in the contract. The absence of that number on the page does not authorize calculating a percentage from sparse observations.
See Laudos.AI in your workflow
Live demo with your exam types and your institution's template, right in the product.