Tools and overview
creating a project
Member missing so reassignment
repository on github

Autonomous Project Manager (Elastic Agent)

This project demonstrates an AI-powered autonomous project intelligence system built using Elasticsearch Serverless, Elastic Agent Builder, ES|QL, Search tools, and MCP integration via ngrok. The system acts as a digital project manager that continuously monitors project state, detects risks, analyzes workload, and generates executive-level insights without requiring manual supervision.

Inspiration

Modern engineering teams rely heavily on dashboards, spreadsheets, and manual check-ins to track project health. These approaches are reactive, fragmented, and inefficient. Critical risks such as deadline slippage, workload imbalance, and stalled tasks often go unnoticed until it is too late.

This project explores a different paradigm: an autonomous agent that continuously observes, reasons, and acts on project data in real time.

Instead of humans querying dashboards, the agent proactively analyzes structured data stored in Elasticsearch and provides actionable intelligence.

This represents a shift from passive monitoring to active autonomous management.

Core Objective

Build a multi-step Elastic Agent that can:

Understand project state from structured memory
Detect risks and anomalies automatically
Reason using ES|QL and Search tools
Generate executive summaries and recommendations
Operate persistently with minimal human intervention

What This Agent Does

Portfolio-level project overview
Detects overdue tasks
Identifies high-risk projects
Analyzes workload distribution
Compares progress vs deadlines
Generates executive summaries
Provides actionable recommendations
Continuously reasons over persistent project memory

Architecture

Brain: Elastic Agent Builder
Memory: Elasticsearch Serverless
Intelligence Layer: ES|QL + Elastic Search tools
Integration Layer: MCP Server exposed securely via ngrok
Interface: Natural language interaction via Agent Builder

MCP Integration and Remote Agent Control

To enable real-time agent interaction and external tool connectivity, the system integrates with the Elastic MCP Server, exposed using ngrok.

This enables:

Remote agent access to Elasticsearch tools
Secure external communication with the agent environment
Future integration with external AI workers and services
Persistent autonomous execution workflows

This transforms the system from a local assistant into a network-accessible autonomous intelligence node.

Data Architecture (Persistent Memory Layer)

The system maintains structured organizational state using Elasticsearch indexes:

team_members — stores skills, workload, and availability
tasks — stores assignments, deadlines, priorities, and required skills
projects — stores project-level context and objectives
ai_workers — stores AI worker capabilities and execution state
activity_logs — stores complete audit trail of agent decisions

This allows the agent to reason over real structured operational data.

Challenges Faced and Solutions

Challenge: Tool integration using MCP server and external connectivity
Solution: Used MCP server exposed via ngrok to securely enable remote tool execution and persistent agent connectivity.

Challenge: Accuracy and repetitive reasoning issues
Solution: Used structured Elasticsearch memory, audit logging, and controlled workflows to ensure consistent and reliable agent behavior.

Challenge: Persistent memory and long-term state management
Solution: Used Elasticsearch Serverless as persistent memory layer for reliable reasoning across sessions.

Future Vision

This system establishes the foundation for a fully autonomous AI workforce where agents continuously monitor, reason, and act without human intervention.

Future capabilities include: