Skip to main content

Public

Project privacy set to public. By default, its content is available to everyone (authenticated or not). Please note that more restrictive permissions might exist on some items.

    •  
      request #41129 Adding an AI Assistant to Facilitate Project Management in Tuleap
    Actions
    Infos
    #41129
    François LE FEVRE (flefevre79)
    2025-01-06 10:53
    2024-12-22 23:43
    42808
    Details
    Adding an AI Assistant to Facilitate Project Management in Tuleap

    Description

    Current Challenge:
    Tuleap users may face difficulties or spend significant time creating tickets, tracking project progress, or applying methodologies like SAFe or Scrum. Standardizing practices, writing complete and relevant tickets, and quickly accessing methodological advice often require expertise that not all users possess.

    Proposed Solution:
    Integrate an AI assistant into Tuleap with the following capabilities:

    1. Ticket Creation and Enhancement:

      • Automatically generate or suggest ticket content based on user input, following best practices for clarity and completeness.
      • Recommend tags, priorities, and categories based on the ticket context.

    2. Project Tracking Support:

      • Provide insights on project progress, detect potential bottlenecks, and suggest actions to resolve issues.
      • Assist in generating status reports or summaries for stakeholders.

    3. Methodology Guidance:

      • Offer contextual advice on implementing SAFe, Scrum, or other methodologies within Tuleap projects.
      • Provide templates, checklists, and best practices tailored to the chosen framework.

    Integration Options:
    The AI assistant could be implemented with flexibility to adapt to various organizational constraints and infrastructures:

    • Use external inference services such as ChatGPT or Gemini for cloud-based solutions.
    • Deploy on-premises inference engines to ensure data privacy and control, leveraging standards like the OpenAI API. For example:
      • Using tools such as Llama 2 or LiteLLM installed within the company’s infrastructure.
    • Support a hybrid approach, allowing organizations to choose between external and internal inference engines based on their needs.

    Expected Benefits:

    • Improve efficiency and reduce time spent on ticket creation and management.
    • Enhance consistency and quality in project tracking and reporting.
    • Support users in applying advanced project management methodologies, improving adoption and adherence.
    • Provide flexibility to organizations to integrate AI while respecting their security and privacy requirements.

    Potential Implementation Details:

    • The assistant could leverage existing AI frameworks, adapting to Tuleap’s workflows and user interface.
    • Integration points include ticket creation forms, dashboards, and project boards.
    • Ensure the assistant respects user permissions and access controls, preserving data security.

    Priority and Value

    This feature would significantly enhance user experience by streamlining workflows, supporting advanced methodologies, and providing flexibility for organizations to adopt AI on their terms. It aligns with Tuleap’s mission to enable efficient and agile project management.

    Roadmap
    Empty
    Empty
    • [x] enhancement
    • [ ] internal improvement
    Empty
    Stage
    Empty
    New
    Empty
    Attachments
    Artifact links
    Empty
    Empty
    References
    References list is empty
    Display settings

    Follow-ups

    User avatar
    Manuel Vacelet (vaceletm)2025-01-06 10:53

    This is a very interesting proposal and even more if there is an opportunity of a joint collaboration on this subject.

    A few key things to keep in mind:

    • Tuleap is very often used in context that don't have access to external services. You already noticed that with the usage of local models, it should also comes with rules of hardware sizing. Very few organization will have the ability to rack GPUs in there datacenter and falling back to CPU usage is likely to come at a cost.
    • Tuleap has a very complex permission model and organizations tend to be very careful with the data they host. Even with local models, we have no guaranty that all the data of a given server could be fed in a single model. I don't know what is the impact here, is that relevant to have a "1 model per project" ? will that be enough data to be interesting ?

    Do you have any inputs on those topics ?