Functional Overview

Issue AI Analytics for Jira introduces new analytics functionality for Jira issues. It exposes an Enable AI Analytics button in the Issue View screen. Once button is pressed, the issue summary and description are sent for analysis to an external AI engine. The analysis results are displayed as tabs, through a new panel in the Issue View screen.

Example:

The tabs that are displayed and the analysis that is done by the AI engine is configurable, depending on the needs of your organization (reference Configuration section).

The following analytics tabs are available by default:

• Instructions – Instructions with proposals and/or concrete solutions on how to solve tasks. It might contain steps to follow, general advise on how to approach the problem or in some cases, a direct solution. This section will become more accurate, as more detailed issue description is provided.
• Description Improvements – Suggestions on how to improve clarity of an issue description. In some cases, the issue author might not provide a description that is detailed or clear enough – this section will help improve the quality and make it more understandable.
• Definition of Done (DoD) – Most teams don’t spend enough time to write Definition of Done (DoD) for issues. This section will automatically generate the DoD. It will help the author to properly define, when an issue can be considered done.
• Test Strategy – Complete strategy on how to test your issues are successfully completed. In most cases, a comprehensive suite of tests will be provided. This might include tests for non-functional requirements or corner cases, which are easily forgotten.

The following AI engines are supported:

• OpenAI GPT-4o
• OpenAI GPT-3.5-Turbo
• Anthropic Claude 3 Opus
• Anthropic Claude 3 Sonnet
• Anthropic Claude 3 Haiku
• Azure OpenAI Service

Architectural Overview

Issue AI Analytics integrates Jira with an external AI provider for analytics through a Heroku server. This can be visualized as follows:

The Deview Studios app server is hosted on the Heroku platform and is the entry point for the integration.

Integration process triggers whenever the Enable AI Analytics button is pressed in the Issue View screen. Only the issue summary and description are sent for analysis.

First time analytics generation for an issue can take between 20-60 seconds. This is primarily due to the processing time needed by the AI provider. Once generated, analytics are cached in an analytics database and every subsequent display of analytics results in the Issue View screen is immediate. Analytics results will be re-generated, if issue summary or description is changed.

The integration process can be visualized as follows

The analytics database only caches analytics results and never stores issue summary or description. Analytics results are stored fully encrypted in the database. Caching is required to provide better user experience, by not mandating the user to wait 20-60 seconds every time.

Requirements

The following conditions must be met in order to use Issue AI Analytics for Jira:

• N/A

Functional Overview

Issue AI Analytics for Jira introduces new analytics functionality for Jira issues. It exposes an Enable AI Analytics button in the Issue View screen. Once button is pressed, the issue summary and description are sent for analysis to an external AI engine. The analysis results are displayed as tabs, through a new panel in the Issue View screen.

Example:

The tabs that are displayed and the analysis that is done by the AI engine is configurable, depending on the needs of your organization (reference Configuration section).

The following analytics tabs are available by default:

• Instructions – Instructions with proposals and/or concrete solutions on how to solve tasks. It might contain steps to follow, general advise on how to approach the problem or in some cases, a direct solution. This section will become more accurate, as more detailed issue description is provided.
• Description Improvements – Suggestions on how to improve clarity of an issue description. In some cases, the issue author might not provide a description that is detailed or clear enough – this section will help improve the quality and make it more understandable.
• Definition of Done (DoD) – Most teams don’t spend enough time to write Definition of Done (DoD) for issues. This section will automatically generate the DoD. It will help the author to properly define, when an issue can be considered done.
• Test Strategy – Complete strategy on how to test your issues are successfully completed. In most cases, a comprehensive suite of tests will be provided. This might include tests for non-functional requirements or corner cases, which are easily forgotten.

The following AI engines are supported:

• OpenAI GPT-4o
• OpenAI GPT-3.5-Turbo
• Anthropic Claude 2.1
• Anthropic Claude 2.0
• Private AI

Architectural Overview

Issue AI Analytics integrates Jira with an external AI provider for analytics through a Heroku server. This can be visualized as follows:

The Deview Studios app server is hosted on the Heroku platform and is the entry point for the integration.

Issue AI Analytics for Jira Data Center also supports Private AI Mode. If enabled, the Heroku cloud server is not used and no data is sent to the Internet, with the expectation that there is a Private AI model hosted within the same datacenter:

Integration process triggers whenever the Enable AI Analytics button is pressed in the Issue View screen. Only the issue summary and description are sent for analysis.

First time analytics generation for an issue can take between 20-60 seconds. This is primarily due to the processing time needed by the AI provider. Once generated, analytics are cached in an analytics database and every subsequent display of analytics results in the Issue View screen is immediate. Analytics results will be re-generated, if issue summary or description is changed.

Requirements

The following conditions must be met in order to use Issue AI Analytics for Jira:

• For Cloud AI mode – connectivity to deview-dc-prod-30130e64aa60.herokuapp.com via HTTP/TCP on port 443 (HTTP proxy is supported)
• For Private AI mode – an internally hosted Private AI server must be available and reachable from Jira

import com.atlassian.jira.util.json.*
def aiQuery = issueSummary + "\r\n" + issueDescription + "\r\n---\r\n" + query // Build the query that will be sent for analysis
def payload = new JSONObject() // Prepare body as per API specification
    .put("model", "gpt-3.5-turbo")
    .put("messages", new JSONArray()
        .put(new JSONObject()
            .put("role", "user")
            .put("content", aiQuery)))
    .put("temperature", 1.0)
    .toString();
def post = new URL("https://api.openai.com/v1/chat/completions").openConnection() // Create a new request to specified URL
logger.debug("Preparing POST body: " + payload)
post.setConnectTimeout(60000) // Set a timeout to 60 seconds - make sure to always set timeouts
post.setReadTimeout(60000) // Set a timeout to 60 seconds - make sure to always set timeouts
post.setRequestMethod("POST") // Specify HTTP method
post.setDoOutput(true) // Specify our HTTP request has a body
post.setRequestProperty("Content-Type", "application/json") // Set HTTP header
post.setRequestProperty("Authorization", "Bearer YOUR_TOKEN") // Set HTTP header
post.getOutputStream().write(payload.getBytes("UTF-8")) // Set HTTP body
def postRC = post.getResponseCode() // Send POST request
def response = post.getInputStream().getText() // Read response body
logger.debug("Response code is: " + postRC + "; Response body is: " + response)
if (postRC.equals(200)) { // Verify request is successful
    def json = new JSONObject(response) // Read as JSON
    def result = json.getJSONArray("choices").getJSONObject(0).getJSONObject("message").get("content") // Traverse JSON tree
    return result // This script must always return the produced analytics
}
throw new Exception("Unsuccessful request! HTTP response code is " + postRC + "; Response body is: " + response) // Throw error due to unsuccessful request