Service Level Agreement

This Service Level Agreement ("SLA") defines the service commitments that Sarudo provides to its clients for the AI Employee platform. This SLA is incorporated into and forms part of the Terms of Service. Capitalized terms not defined herein have the meanings set forth in the Terms of Service.

1. Service Availability

Sarudo does not currently offer a contractual uptime percentage. We would rather state that plainly than publish a figure we cannot yet evidence.

What we do commit to: your instance runs on infrastructure dedicated to you, and we restore service when it is interrupted by something within our control. "Downtime" means any period during which your AI employee cannot receive or respond to messages across its configured channels for reasons within Sarudo's control. Scheduled maintenance and the exclusions in Section 8 are not downtime.

Interruptions caused by credentials you provision — an expired API key, an exhausted quota, a declined card, a lapsed OAuth grant — are outside Sarudo's control. We will tell you when we detect one, but resolving it requires action on your account.

2. Scheduled Maintenance

Sarudo performs regular maintenance to ensure optimal performance, security, and reliability of the platform. Our maintenance commitments include:

• Advance Notice: We will provide at least 48 hours' notice for scheduled maintenance that may affect service availability. Emergency security patches may be applied with shorter notice when necessary.
• Maintenance Windows: Scheduled maintenance is performed during low-traffic periods, typically between 2:00 AM and 6:00 AM in the client's primary time zone, unless otherwise agreed.
• Duration: We aim to complete scheduled maintenance within 2 hours. If maintenance is expected to exceed this window, we will communicate the expected duration in advance.
• Notification: Maintenance notifications are sent via email to the primary contact associated with your account and, where applicable, via your configured communication channels.

3. Incident Response

Sarudo classifies incidents by severity and provides corresponding response commitments:

Severity 1 — Critical

Definition: AI employee is completely unavailable or unable to process any communications.

• Response Time: Within 24 working hours at the outside — the same ceiling as every other severity, because a total outage must never be answered more slowly than a cosmetic bug. Within business hours, the same day, and ahead of everything else in the queue.
• Update Frequency: Until resolved, at least once per business day
• Resolution Target: We do not publish one. Severity 1 takes priority over all other work until service is restored.

Severity 2 — High

Definition: AI employee is operational but with significantly degraded performance or partial functionality loss (e.g., one communication channel is down).

Severity 3 — Medium

Definition: Minor issues that do not significantly affect the AI employee's core functionality (e.g., slower response times, minor formatting issues).

Severity 4 — Low

Definition: Feature requests, cosmetic issues, or minor improvements.

What we commit to on Severity 2, 3 and 4

One response commitment: within 24 hours, counted on working days. Not four hours for one tier and forty-eight for another — a single number, for every ticket that is not a Severity 1. It is the time until a human responds to you, not the time until the issue is fixed.

The clock only runs Monday to Friday. We do not operate a weekend or out-of-hours rota, so we do not pretend the clock does. In practice: a ticket raised Tuesday morning is due Wednesday morning, and one raised late on Friday is due Monday — not Saturday. Indian national holidays are excluded on the same basis. We would rather you could predict that from this page than discover it on a weekend.

We do not publish a resolution target. Severity governs the order we work in, not a promised finish date: Severity 1 comes before everything else, then 2, then 3, then 4. How long a fix takes depends on what broke, and quoting a figure we cannot evidence is how the previous version of this page came to carry four separate resolution promises that nothing measured.

This commitment is measured, not aspirational — for anything that becomes a ticket. Raising one through your AI employee stamps it with a working-day target on creation, and stamps it again if that target is missed, so our record against it is a fact we hold rather than a claim we make, and you are entitled to ask us what it looks like. Email is the exception: writing to us does not currently open a ticket, so nothing clocks it. If you want the commitment to apply, raise it through your AI employee — and until we close that gap, we would rather tell you than let you assume otherwise.

4. Support Channels

Sarudo provides support through the following channels:

• Email: hello@sarudo.com — monitored during the support hours below. Response times are the ones in Section 3; this channel does not carry a separate or faster commitment.
• Instant Messaging: Direct messaging via a dedicated support channel for real-time communication during support hours.

Support hours are Monday through Friday, 9:00 AM to 6:00 PM IST, excluding Indian national holidays. We do not currently operate a 24/7 on-call rota, and we will not claim one. If your instance stops responding outside those hours, report it through any support channel — it is picked up at the start of the next business day, or sooner if someone is available.

5. Performance Standards

Sarudo commits to the following performance standards for your AI employee:

• Responsiveness: Your AI employee acknowledges a message and begins work on it as soon as it is received. We do not publish a numeric latency target, because response time depends on the model provider you hold the account with and on what the request requires — a one-line reply and a multi-step task that reads three documents are not the same job. We will not commit to a number we do not measure.
• Capability Reliability: Configured capabilities function as documented. Where one depends on an account you hold, its availability depends on that account remaining funded and authorised — see Section 1. We do not publish a percentage reliability target per capability.
• Knowledge Base Accuracy: The AI employee will draw from and accurately reference your provided knowledge base materials. Accuracy is continuously improved through feedback and training refinements.

6. Data Backup

Sarudo implements comprehensive data backup procedures to protect your information:

• Frequency: Automated daily backups of your entire AI employee configuration, knowledge base, and conversation history.
• Retention: Backups are retained for 30 days on a rolling basis.
• Storage: Backups are written to your own server — the same host the instance runs on. Each one is verified as readable after it is written and recorded with a SHA-256 checksum, and they are stored with owner-only permissions because the archive includes your credentials. This protects against accidental deletion and corruption. It does not protect against losing the host itself. If you want copies held somewhere else, we will configure that to a destination you own and control — it is not enabled by default, and we will not describe it as off-site until it is.
• Recovery: Backups are taken daily and verified, so the most you would lose is a day's work. We do not publish a restore-time target: we have not rehearsed a full restore often enough to promise a number, and we would rather say that than quote one. If you need a committed recovery time, ask and we will rehearse a restore on your instance and agree it in writing.

7. Service Credits

Sarudo does not currently operate a service-credit scheme, because it has no published uptime percentage to measure against (see Section 1). Owning your instance carries no recurring fee. The optional Heartbeat plan is the one recurring charge, and it buys a fixed allocation of a person's time rather than an availability guarantee — so there is no uptime commitment attached to it to credit against either. You may drop it at any time.

If an interruption within Sarudo's control materially affects your instance, raise it through any support channel. We will tell you what happened, what we did, and what we changed so it does not recur. Where an interruption consumed support hours you had already purchased, those hours are returned to your balance.

8. Exclusions

The availability commitments in Section 1 do not apply to downtime or performance issues caused by:

• Force Majeure: Events beyond Sarudo's reasonable control, including natural disasters, wars, pandemics, government actions, or widespread internet outages.
• Client-Side Issues: Problems with your own systems, networks, or configurations that affect your ability to interact with the AI employee.
• Third-Party Service Outages: Outages or disruptions in third-party services integrated with your AI employee (e.g., Twilio for voice/SMS, email providers, CRM systems, messaging platforms) that are outside Sarudo's control.
• Scheduled Maintenance: Pre-announced maintenance windows as described in Section 2.
• Client Requests: Downtime resulting from changes or actions requested by you, including configuration modifications, knowledge base updates requiring reprocessing, or requested restarts.
• Abuse or Misuse: Service disruptions caused by your violation of the Acceptable Use Policy or Terms of Service.

9. Contact

For SLA-related inquiries or incident reports, please contact us at:

• Email: hello@sarudo.com