Service Level Agreement.

This Service Level Agreement ("SLA") describes what subscribers can expect from NarrateEMS with respect to service availability, support, and incident communication. It applies to all active subscriptions and is incorporated by reference into the applicable service agreement.

We've written this to be honest about where we are as an early-stage product. We'll update it as we grow.

1. Service availability

NarrateEMS is a Chrome extension that relies on the following underlying infrastructure:

• Supabase (PostgreSQL + Edge Functions)
• Microsoft Azure OpenAI
• Groq API

NarrateEMS does not independently guarantee a specific uptime percentage. Service availability is dependent on the uptime of these underlying providers, each of which publishes their own SLAs:

NarrateEMS will make commercially reasonable efforts to ensure the extension and backend infrastructure remain operational. In practice, the vast majority of outages will originate from one of the above providers, not from NarrateEMS-specific code or infrastructure.

What this means in plain terms: if Supabase or Azure goes down, NarrateEMS goes down with it. We don't have redundant fallback infrastructure at this stage. We'll be transparent about incidents as they happen.

2. Planned maintenance

NarrateEMS may perform planned maintenance on backend infrastructure from time to time. We will provide at least 48 hours advance notice by email to the subscriber point of contact before any maintenance that is expected to cause a service interruption.

Where possible, maintenance will be scheduled during overnight hours (11pm–5am ET) to minimize impact on active shifts.

3. Incident communication

In the event of an unplanned service outage or degradation that affects the ability to generate documentation:

• NarrateEMS will post a status update to listed contacts within 2 hours of confirming the incident.
• We will provide updates at regular intervals until the issue is resolved.
• A brief post-incident summary will be sent within 3 business days of resolution, describing the cause and steps taken to prevent recurrence.

For real-time status, subscribers can check the status pages of our infrastructure providers listed in Section 1.

4. Support

How to reach us

Email: narrateems@gmail.com

All support requests should include: your account email, a description of the issue, and the device and browser version you're using.

Response times

NarrateEMS is an early-stage company operated by a small founding team. We commit to the following on a best-effort basis:

"Best effort" means we have no contractual penalty for missing these targets at this stage. We include them because we want to be transparent about our intentions, not because we're making a legally enforceable promise. As we grow, this SLA will be updated with binding commitments.

Support hours

We don't currently offer 24/7 support. Support is monitored during standard business hours, Monday through Friday, 9am–6pm ET. We'll do our best to respond to urgent outage reports outside those hours when we see them, but cannot guarantee it.

5. Remedies

NarrateEMS does not currently offer service credits or refunds for downtime. If you experience a significant and prolonged outage that materially impacts your operations, contact us at narrateems@gmail.com and we'll work something out on a case-by-case basis. We'd rather make it right than lose a customer over something outside our control.

This will be replaced with a formal credit schedule in a future version of this SLA.

6. Session failure and connectivity — expected behavior

The following behaviors are by design and do not constitute service failures under this SLA:

• Connectivity loss mid-session: If a device loses internet connectivity after a narrative has been transcribed but before the documentation generation request completes, the transcript is held in memory and retried up to 3 times with a 2-second delay. If all retries fail, the session fails and the transcript is not recoverable. The user must re-dictate. There is no persistent retry queue that survives a browser restart or extension close.
• Groq unavailability: If the Groq page relevance service is unreachable, the extension degrades gracefully and processes the current EMS Charts page only. Documentation generation via Azure OpenAI continues.
• Azure OpenAI unavailability: If the field extraction service is unreachable, the extension retries 3 times. If all retries fail, the session fails. No cross-provider fallback exists.
• Shared device session state: Session data in chrome.storage.local is cleared only on manual reset. On shared devices, users should reset their session after each use. Residual session state from a prior user is not a service failure.

In all session failure scenarios, the user's underlying EMS Charts data is unaffected — NarrateEMS only writes to form fields that the user explicitly approves. A failed NarrateEMS session never corrupts or alters existing patient records.

7. Exclusions

This SLA does not apply to service unavailability caused by:

• Factors outside NarrateEMS's reasonable control, including upstream provider outages (Supabase, Azure OpenAI, Groq), internet service disruptions, or acts of nature.
• Actions taken by the subscriber, including misconfiguration of devices, unauthorized use, or use in violation of the Terms of Service.
• Scheduled maintenance communicated in advance per Section 2.
• Session failures caused by device connectivity loss as described in Section 6.
• The free trial period.

8. Updates to this SLA

NarrateEMS will update this SLA as the product matures, infrastructure improves, and the team grows. Material changes will be communicated by email at least 14 days before taking effect.

The most current version will always be available at narrateems.com/sla.

Contact

NarrateEMS
Email: narrateems@gmail.com
Website: www.narrateems.com

© 2026 NarrateEMS Inc. All rights reserved.