Support SLA Definitions and Examples

Knowledge Articles
1

“Understanding Severity Levels: Categorizing Platform Issues Based on Impact”

This article offers clarification and descriptions of severity levels for issues reported on the platform, categorizing them based on their impact.
NOTE - Severity should not be mistaken for priority. For instance, a customer may consider a question urgent, resulting in a Sev4 severity, while the customer priority remains high.

Severity Definitions:

Severity-1 - Critical

A “Critical” or “Severity 1” Error renders the Platform/Bot Support completely unusable or nearly unusable or

introduces a high degree of operational risk. No Workaround is available. Until this Error is resolved, the Platform’s/Bot’s use is essentially halted. A large number of users and/or Kore Platform/Bot functionality is

severely impacted. Any Error that is faced in the course of the usage of any functionality reserved or restricted for

Administrator/Bot Developer shall be excluded from the aforementioned definition.

Examples -

  • XO Platform
  1. Unable to log into the Platform from any network
  2. Unable to get a response from a published VA more than 80% of the time.
  3. Key component failures (applicable for on-premise installations predominantly) like Redis, MongoDB, or RabbitMQ down.
  • XO Contact Center AI (SmartAssist)
  1. Agents / Supervisors unable to login
  2. No Agent transfer scenario is working.
  3. Agent Console / Desktop becomes unresponsive
  4. Supervisor Console becomes unresponsive
  5. Over 10% of the calls are dropping
  • AgentAssist
  1. AgentAssist not responding to any utterances
  2. Library tasks not displayed or not responding
  • SearchAssist
  1. Unable to log into the Platform from any network
  2. Unable to get a response from a SearchAssist application more than 80% of the time
  3. SearchAssist responses not coming in.

Severity-2 - High

A “High” or “Severity 2” Error renders one or more critical functionalities of the Platform/Bot are unavailable;

however, the overall performance of the Platform/Bot is not impacted. Workarounds may be available but use of the

The platform/Bot is degraded and causes continuing operational risk. A moderate number of users are significantly impacted, but overall the Platform/Bot Support continues to function.

Examples -

  • XO Platform
  1. Agent transfer not being initiated.
  2. The end user is unable to get a response from a published VA between 20-80% of the time.
  3. VA is unable to recognize most of the intents apart from small talk.
  4. ML/FAQ training fails for all the bots on the Platform.
  5. Delayed responses from the bot that take more than 8 seconds.
  6. All (or over 80%) of the API calls (Service nodes) are failing but on tools like Postman, no failures are observed.
  7. Login MFA OTPs or email verification emails not going out on a region or server level (Not specific to a workspace or user - User specific issues are mostly due to their organization policies and email server rule specific)
  • XO Contact Center AI (SmartAssist)
  1. Agent transfer is working but agents on Kore.ai agent desktop are not able to accept more than 60% of the chats.
  2. Over 5-10% of the calls are dropping
  3. Delay in response to the customer
  4. Incidents related to Skills / Queues / Routing of requests not working
  5. The supervisor Console not refreshing the page frequently
  6. Analytics or dashboards are not loading even for a 24-hour date range
  • AgentAssist
  1. Send / Copy buttons not working
  2. AssistAssist search is not working
  3. AgentAssist metrics/dashboards not loading for even 24 hours range
  • SearchAssist
  1. User is unable to get response from searchassist 50-80% time
  2. Search query without generative AI enabled taking more than 8 sec.

Severity-3 - Medium

A “Medium” or “Severity 3” Error is an inconvenience or causes inconsistent behaviour, which does not impede

the normal functioning of the Platform/Bot. It could be an Error that occurs inconsistently and affects

nonessential functions or is an inconvenience that impacts a small number of users. It may also contain

visual errors where the graphical display of the Platform/Bot is not ideal, but still functioning correctly.

Examples -

  • XO Platform
  1. Unable to publish a VA (VA specific) and no apparent reason is displayed on the product.
  2. Unable to import a VA onto a target VA (from an environment running on a supported version lesser than or equal to the target environment version).
  3. VA specific export issues.
  4. The end user is unable to get a response from a published VA between 5-20% of the time.
  5. The platform is unable to initiate Agent transfer intermittently for off-the-shelf integrations offered by the Platform.
  6. The platform is unable to send out emails (non-login related)
  7. Analytics and dashboards are not loading for a certain date range.
  8. Delay in VA response for over 3 seconds up to 8 seconds without the involvement of any third-party API calls/intervention in between.
  9. Builder tool becomes unresponsive
  10. Unable to switch workspaces
  • XO Contact Center AI (SmartAssist)
  1. A particular flow component is not working as expected but the business of end-users is not severely impacted.
  2. Sporadically calls are dropping.
  3. Unable to publish the flow.
  4. Analytics/dashboards are not loading for some specific date range over 24 hours.
  5. Transcripts are not working in specific scenarios
  6. Some of the pages are not accessible or are becoming unresponsive to the Supervisor
  7. Configuration changes/settings are not reflecting as intended
  8. Batch test inaccuracies or discrepancies.
  • AgentAssist
  1. AgentAssist widget settings not working
  2. Switching of Layouts not working
  • SearchAssist
  1. Web-crawler or scheduled ingestion service taking for than 1 hour per 100 documents
  2. App training is failing sporadically
  3. Search calls are failing 30-50 % time.

Severity-4 - Low

A “Low” or “Severity 4” Error has a small degree of significance, is a minor cosmetic issue, or is a “one-off”

case. A “one-off” case occurs when the Error occurs infrequently and cannot be reproduced easily. These are

Errors that do not impact the daily use of the Platform/Bot. A Low Error is something that does not affect normal use, and can be accepted for a period of time, but the user would eventually want to be changed.

Examples -

  • XO Platform
  1. Intermittent issues or incidents with any Kore product that replicate less than 5% of the time for the end-users of the Kore.ai product.
  2. A cosmetic UI issue.
  3. A one-off issue.
  4. An issue for which no known steps to replicate exists.
  • XO Contact Center AI (SmartAssist)
  1. Language-specific responses not coming intermittently.
  2. Minor UI-specific issues with no functional impact
  • AgentAssist
  1. Minor UI-specific issues with no functional impact
  2. Language-specific responses not coming intermittently.
  • SearchAssist
  1. Builder UI issues.
  2. Rarely some search calls fail.
  3. Generative AI results are intermittently failing.
  • Common
  1. Cosmetic defects.
  2. One-off cases or cases that can be rarely replicated
  3. Feature/ Enhancement Requests
  4. Specific Documentation related queries or bugs.
  5. How to ?- Questions
  6. Adhoc requests like a particular script/access to support portal/bot-kit deployment requests.
  7. Provisioning related requests - provisioning a new server/environment for load testing, etc.
  8. Root Cause Analysis
    • Issue not currently occurring/happened in the past - request is to know the root cause
    • Follow up ticket after resolution of an issue to know the root cause.