← Back to Posts

Iradaty Bot (Case Study)

Overview

Iradaty Bot is an AI-assisted reporting and knowledge workflow built to reduce the operational burden on a small humanitarian NGO team working under high uncertainty and limited capacity. It turns high-frequency, messy updates into structured reports that leadership can act on.

The problem

The team did report; but the information arrived scattered across messages, screenshots, and voice notes. Weekly reporting consumed leadership time and still missed patterns. Institutional memory stayed fragmented; and past decisions were hard to retrieve.

The design choice

I did not force a new tool first. The system meets the team where they already are (chat-based updates); then adds structure through automation.

What the workflow does

  • Collects periodic updates (field + admin)
  • Produces standardized outputs:
    • Personal weekly reports
    • Department reports
    • A management weekly report
  • Stores outputs in Notion so they are searchable, reviewable, and reusable

Architecture (high level)

  1. Input layer: chat-based reporting
  2. Processing layer: LLM summarization and structured extraction
  3. Storage layer: Notion (raw inputs + outputs + knowledge base)
  4. Output layer: scheduled reports and ad-hoc summaries

Impact

  • Reduced manual effort required for consistent weekly reporting
  • Improved visibility across departments through standardized outputs
  • Built a searchable archive of decisions, blockers, and recurring themes

Lessons

  • In low-capacity environments, automating around habits creates leverage.
  • “AI summaries” become operationally useful when stored in a system built for search, history, and reuse.