AI Project Assistant (code: LAB25001)

Executive Summary

The AI Project Assistant is designed to support project management at M31 by providing tools for advanced project data analysis and assistance in creating technical and operational documentation. In addition to monitoring and interpreting information from various sources, the assistant will guide teams in drafting formal and operational documents, generating templates and pre-structured stubs for documents compliant with procedures (such as ISO9001) and detailed technical documentation.

This project aims to develop an intuitive web application with analysis and suggestion functionalities, as well as automated documentation models to enhance efficiency and accuracy. It will integrate with tools already in use, such as Confluence and Jira. Expected outputs include data analysis dashboards, a notification system for alerts, and a template generator for project documentation.

Estimated costs cover server and software licenses with an initial budget of €500-€1000. The main risks involve the accuracy of data analysis models and the compliance of generated documents with standards, requiring thorough testing to ensure quality.

Problem Description

Managing multidisciplinary and long-term projects, such as those carried out at M31, presents high complexity. These projects require multiple tools for activity tracking, resource management, and data analysis, each generating and storing information in different formats. This fragmentation leads to information dispersion, making it difficult to obtain an overall view and access relevant data, thus reducing the efficiency of the entire management process.

Another critical aspect is the need to produce accurate and consistent documentation. “Internal” documentation is essential for tracking project decisions, tests, analyses, and architectures in detail, providing a solid foundation for the team’s work and future project updates. Simultaneously, “external” documentation must meet regulatory and certification requirements, ensuring compliance with industry standards (e.g., ISO9001). Writing this documentation is often seen as a burdensome task, leading to it being neglected or completed hastily, which can result in informational gaps and compromise project quality and compliance.

The Intelligent Project Assistant is proposed as a solution to simplify documentation creation and management by providing the project team with templates and documentation models for both internal use and external compliance standards. At the same time, the assistant analyzes collected data, helping the project manager monitor overall progress, identify risks and potential issues, and take timely corrective actions, thereby improving project oversight and management effectiveness.

Specific Objectives and KPIs

Specific Objectives:

Provide the team with an intuitive system to create accurate and compliant project documentation, reducing time and errors in drafting official and technical documents.
Centralize relevant project information to improve traceability and facilitate data access for all team members.
Enhance project monitoring with updated dashboards highlighting progress, potential issues, and emerging risks, providing decision support to the project manager.
Integrate the assistant with tools currently in use at M31 (Bytes, Jira, Confluence, Bitbucket, Teams, and SharePoint).

Key Performance Indicators (KPIs):

Documentation completion time: Reduce the time required for drafting and reviewing project documentation by 30%, measured in man-hours compared to current timelines.
Documentation compliance rate: Achieve at least 95% compliance for documentation required by ISO9001 procedures and quality audits.
Project information coverage: Ensure that at least 80% of key project decisions and tests are documented and traceable within the first three months.
Team satisfaction: Target a team satisfaction rating of ≥4 out of 5, monitored through quarterly surveys to assess the assistant’s usefulness and usability in document and decision-making support.

Feasibility Analysis

From a technical perspective, the project is feasible as all necessary technologies are available. These primarily include frameworks for creating AI agents (LangGraph), cloud execution environments (Google Cloud Platform), RAG implementation techniques (LangChain), and HTTP APIs.

Potential risks lie in ensuring that the agents meet the specified KPIs and produce predictably reproducible results. This risk is inherent in LLM-based agents, which are statistical and non-deterministic models. To mitigate this risk, best practice techniques will be implemented to guide and control the output generated by predictive models.

High-Level Architecture and Proposed Technologies

A cloud-native architecture is proposed, based on a backend-frontend separation, with the frontend being web-based. The backend is planned as a microservices infrastructure (potentially serverless).

Proposed technologies (non-exhaustive list):

Python/Go: for programming the AI agents
Go/TypeScript: for backend and frontend service implementation
Angular: as the UI framework for the web frontend
LangChain/LangGraph: for AI agent creation and management
PostgreSQL/MongoDB: as databases

Project Phases and Work Plan

The project is structured into the following phases:

Phase 1: Analysis of required functionalities and creation of the basic structure of assistants integrated with existing tools.
Phase 2: Implementation of task monitoring and suggestion logic.
Phase 3: Development of system infrastructure and graphical interface.
Phase 4: Testing and optimization based on user feedback.
Phase 5: Release and documentation.

The preliminary timeline (to be refined during the first month of the project) is as follows:

Phase 1: Months 1-3
Phase 2: Months 3-5
Phase 3: Months 1-5
Phase 4: Months 5-7
Phase 5: Month 8

Expected Value and Impact

The project will significantly contribute to M31’s project management approach by creating a tool that enhances project oversight. The intelligent assistant will enable students to learn how to develop AI-driven productivity solutions and could serve as a foundation for other project management applications at M31 and in the broader adoption of GenAI solutions.

Required Resources and Estimated Budget

Logistics: A workspace for a team (ideally 3-4 people).

Software: Accounts for accessing LLM HTTP APIs (e.g., OpenAI).

Hardware: Access to virtual machines and infrastructure for AI tasks and application execution.

Estimated Budget:

APIs: €80/month → €640 (over 8 months)
Virtual machines and cloud environment: €150/month → €1200 (over 8 months)