Office Memory AI-Never Lose the "Why" Behind Team Decisions.

Office Memory AI helps teams remember not just what was decided in Slack, but why. It captures important decisions, preserves their reasoning, and provides intelligent, context-aware answers.

Comment 1

Inspiration

Every organization makes hundreds of important decisions inside Slack. Over time, those decisions become buried under thousands of messages, making it difficult for employees to understand why a particular technology, policy, or process was chosen. New team members spend hours searching through old conversations, and valuable institutional knowledge is often lost when employees leave.

We built OfficeMemoryAI to solve this problem. Instead of simply searching Slack, it remembers the reasoning behind important decisions, preserves organizational knowledge, detects conflicting decisions, and instantly answers questions with source-backed explanations.

What it does

OfficeMemoryAI is an AI-powered Slack agent that automatically captures important team decisions from conversations. OfficeMemoryAI uses Slack's native AI capabilities through the Agents & Apps experience, allowing users to ask natural-language questions directly inside Slack. The agent retrieves relevant organizational decisions, explains the reasoning behind them, highlights rejected alternatives, and provides context-aware answers from previously captured team discussions.

It can:

  • Automatically detect decision messages in Slack.
  • Extract the final decision, reasoning, and rejected alternatives using AI.
  • Store important decisions as long-term organizational memory.
  • Answer questions like "Why did we choose PostgreSQL?" using previous discussions.
  • Detect contradictory decisions before they create confusion.
  • Ignore casual conversations that are not actual team decisions.
  • Preserve organizational knowledge so new employees never lose historical context.

Instead of forcing employees to search through hundreds of Slack messages, OfficeMemoryAI provides instant, explainable answers backed by previous team discussions.

How we built it

Our AI memory pipeline works as follows:

Slack Message
      │
      ▼
Decision Detection
      │
      ▼
AI Decision Extraction
(Decision • Reason • Rejected Alternatives)
      │
      ▼
Contradiction Detection
      │
      ▼
Persistent Memory Storage
      │
      ▼
AI-powered Question Answering
      │
      ▼
Slack Response

Technology Stack

  • Python
  • Slack Bolt
  • OpenAI-compatible API (Qwen)
  • Railway
  • JSON-based Persistent Memory
  • AI-powered Decision Retrieval

Challenges we ran into

  • Identifying which Slack messages actually represent decisions.
  • Extracting structured reasoning from natural language conversations.
  • Preventing hallucinations while answering historical questions.
  • Detecting contradictory decisions across multiple discussions.
  • Designing a lightweight persistent memory system without requiring complex infrastructure.
  • Ignoring casual conversations while preserving meaningful organizational knowledge.

Accomplishments that we're proud of

  • Built a fully functional AI-powered Slack agent from scratch.
  • Automatic decision detection and structured AI extraction.
  • Long-term decision memory that survives conversations.
  • AI-generated reason-aware answers with contextual explanations.
  • Automatic contradiction detection between historical decisions.
  • Lightweight architecture that can easily scale and integrate with enterprise workflows.

What we learned

During this hackathon we learned:

  • Building production-ready Slack Agents using Slack Bolt.
  • Designing long-term AI memory architectures.
  • Prompt engineering for structured information extraction.
  • Retrieval-Augmented Generation (RAG).
  • Building AI systems that preserve organizational knowledge instead of functioning as simple chatbots.
  • Deploying AI applications on Railway.

What's next for OfficeMemoryAI

Our roadmap includes:

  • Multi-workspace support.
  • Microsoft Teams integration.
  • Google Workspace integration.
  • Meeting transcript ingestion.
  • Automatic decision timelines.
  • Decision approval workflows.
  • Rich analytics dashboard.
  • Enterprise-grade RBAC.
  • Knowledge graph visualization.
  • Email and Jira integration.

Features

  • Automatic Decision Detection
  • AI Decision Extraction
  • Decision Reasoning
  • Rejected Alternatives Extraction
  • Persistent Organizational Memory
  • AI-powered Slack Agent
  • Context-aware Question Answering
  • Organizational Knowledge Preservation
  • Proactive Contradiction Detection
  • Enterprise-ready Architecture

My Contribution

I independently designed and built OfficeMemoryAI from scratch, including the Slack bot, backend architecture, AI-powered decision detection, structured decision extraction, contradiction detection, persistent memory system, Railway deployment, prompt engineering, demo video, documentation, and the complete Devpost submission.

Built With

Share this project:

Updates

posted an update —

Hello Judges,

Thank you for the feedback.

I have updated the submission to clearly demonstrate the required Slack technologies.

Changes made: • Added a dedicated demo showing Slack's native Agents & Apps (Slack AI capabilities). • Updated the YouTube demo link with the latest version demonstrating the AI Agent workflow. • Expanded the project description to explicitly explain the use of Slack AI capabilities. • Added the Slack AI Agent demo link for easier verification. • Integrated MCP server support and documented it in the submission. • Updated the architecture and implementation details to better reflect the current project. • Verified the deployed application and shared workspace access for testing.

The Slack workspace access has already been provided for evaluation.

Thank you for reviewing OfficeMemoryAI.

Log in or sign up for Devpost to join the conversation.