Integrate Artificial Intelligence (AI) to Confluence Data Center

[#1]

In 2023, Atlassian introduced the Intelligence feature, a significant advancement designed to enhance productivity and streamline workflows. This new functionality has been widely acclaimed for its ability to provide smarter, more efficient solutions to users. Nevertheless the Intelligence feature is available exclusively for Atlassian Cloud products. Many organizations using Confluence Data Center are eager to benefit from these innovations. How can the Intelligence feature be integrated into Confluence Data Center?

AI for Confluence is just the one to bring intelligence features to Confluence Data Center.

Prerequisite of Using AI for Confluence

To utilize AI in Confluence, the following steps are necessary：

a) Install App

Please download the AI for Confluence app from the Atlassian Marketplace and install it in your Confluence environment. Note that a valid license is required for the AI features to function properly; without it, the AI capabilities will not be operational.

b) Connect a Large Language Model

Using AI for Confluence requires the presence of a Large Language Model (LLM), the App supports multi LLMs including:

• Commerical Large Language Model

  • OpenAI ChatGPT, Azure openAI, QWen

• Open source LLM deployed on primise

  • Models based on the Ollama framework:

    • Llama, Gemma, Mistra, Phi, …

It’s required for Confluence administer to configure AI Provider in the app settings first.

c) Configure Knowledge Source

The Knowledge Source is the data repository from which the AI draws information when utilizing AI features in Confluence. It provides the context for AI interactions, supplying the foundational information that shapes its responses and functionality.

Confluence administrators can configure the Knowledge Source by adding one or more sources. They can also define the scope of each source, including specifying which spaces or CQL queries are included.

In addition to Confluence pages, attachments can also be part of the Knowledge Source. AI for Confluence will extract content from these attachments to serve as a source of information for its interactions. Commonly parsed attachments include:

• Office docs (.docx, .pptx, .xlsx)

• PDF (.pdf)

• Text files (.txt, .xml, .html, .rtf etc)

d) Setup Vector DB (not mandatory)

By default, AI for Confluence includes an in-memory Vector Database that is ready to use without any additional configuration. Additionally, you have the option to deploy a dedicated Vector Database if needed for enhanced performance or specific requirements.

It’s recommended that in-memory database is only used for testing or demo environments. For production environments, it is suggested to use Chroma DB.

In Memory

Due to limitations in memory size, not all data can be stored in the in-memory database. The system includes a configuration parameter Max Index Number to set the maximum number of entries.

Chroma DB

Chroma DB is an open-source tool recommended for use as a database if Confluence has a large volume of data to index.

How AI for Confluence Works

AI for Confluence will combine content from Knowledge Source as the context for the query provided to the LLM when user queries. The LLM will generate responses based on the query and context, and we will provide the answers to the users along with the source links.

Highlight Features of AI for Confluence

Chat with Confluence

Users can initiate conversations directly within Confluence, similar to using ChatGPT. However, AI for Confluence distinguishes itself by leveraging the content from Confluence’s Knowledge Source—including pages and attachment documents—combined with large language models (LLMs) to provide informed and relevant answers. Additionally, it will reference the specific sources within Confluence where the information originated.

Demo Video: Chat with Confluence

Intelligent Query

You can use the ASK AI feature directly from the search bar in Confluence to pose questions to the AI. Unlike Confluence's default keyword-based search, AI for Confluence offers semantic search capabilities. This means that the AI will understand the user's query, integrate the Knowledge Source content with the large language model (LLM), and provide answers. Additionally, it will cite the sources of the answers within Confluence.

Demo Video: Intelligent Query

Content Generation in Confluence Editor

Similar to Confluence Cloud or Notion, when editing a Confluence page, users can leverage AI to assist with writing and enhance productivity. These functionalities primarily include:

• Summarize writing

• improve writing

• Fix spelling & grammer

• Brainstorm

• Make shorter

• Change tones to:

  • professional

  • empathetic

  • casual

  • neutral

  • educational

Demo Video: Content Generation

Summarize Page and Text

When browsing Confluence pages, you can use AI to summarize the entire page or selected excerpts. This feature helps users quickly grasp the content within Confluence, providing concise overviews and highlighting key points for efficient understanding.

Demo Video: Summarize

Define Terms by AI

One of the significant challenges in consuming content within Confluence is obtaining the necessary context to fully comprehend the material. Abbreviations, acronyms, unfamiliar terminology, nomenclature often necessitate extensive searches for additional information. AI for Confluence addresses this issue by providing definitions and explanations directly. Additionally, the sources of these definitions are included, ensuring transparency and reliability.

Demo Video: Define Terms