AI Memory: Impact Report Portal - Apr 27, 2026

  1. Overview
  2. AI
  3. AI Memory Notes
  4. AI Memory: Impact Report Portal - Apr 27, 2026

To create a high-fidelity "Open Brain" memory for your AI agent (compatible with systems like OpenBrain, Claude Code, or custom MCP setups), we must structure the data into Axioms (unchanging truths), Project Context (goals and stakeholders), and Technical Schemas (the blueprint).

Below is the optimized memory block. You can copy and paste this into your AI's persistent memory file (e.g., CLAUDE.md, memory.json, or a Supabase thoughts table).

## 🧠 AI PROJECT MEMORY: HCUSA Impact Portal

1. CORE IDENTITY (Axioms)

  • Organization: Human Concern USA (501(c)3 Nonprofit).

  • Mission: Data-driven humanitarian transparency.

  • Primary Goal: Transition from manual Word-based reporting to a streamlined "Intake → Vet → Publish" pipeline.

  • Stakeholders: Fazil Hussain (Client Rep), Partners (Data Providers), Backend Staff (Vetters), Donors (End Users).

2. PROJECT CONTEXT & CONSTRAINTS

  • Current Bottleneck: Partners fill out Word docs; data entry is manual and inconsistent.

  • Desired Workflow: Google Forms (Partner Intake) → Google Sheets (Staff Vetting/Editing) → Web Portal (Public Display).

  • Key Requirement: The portal must support bulk-uploading of vetted data via Excel/CSV to eliminate manual copy-pasting.

3. TECHNICAL ARCHITECTURE (Hybrid Specification)

  • Backend (The Engine): Laravel (PHP 8.3)

    • Why: Superior for Excel/CSV parsing and complex Role-Based Access Control (RBAC).

  • Frontend (The Face): Next.js (React)

    • Why: Essential for SEO (donor discovery), speed, and modern interactive maps.

  • Database: PostgreSQL (for relational data integrity).

  • Media: Cloudinary or AWS S3 for field photo/video hosting (do not store on-server).

  • Admin Tools: Shadcn UI for a professional, accessible internal dashboard.

4. DATA SCHEMA (The "Impact" Object)

  • Identity: Project ID, Sector (Water, Food, etc.), Region/Country.

  • Metrics: Lives Impacted (Total), Budget, Timeline.

  • Demographics: Children, Women, Men (Integer counts).

  • Deliverables: Specific units (e.g., "50 Hand Pumps", "100 Food Kits").

  • Media: Field photos, videos, implementation logs.

5. RISK & DECISION LOG

  • Decision: Rejected a pure manual-entry CMS in favor of an Excel-to-Web integration.

  • Risk: Data formatting errors during Excel upload.

  • Mitigation: Implement strict backend validation (Mirkadim Proposal) to reject rows with missing mandatory fields.

  • Accountability: Vendor is Progressive Byte (Sabbir Rahman) for professional agency-grade code, or Mirkadim for high-efficiency solo delivery.

6. INSTRUCTIONS FOR THE AI AGENT

  • Always prioritize: Scalability for 5+ years and ease of use for non-technical staff.

  • Tone: Professional, nonprofit-aligned, and technically precise.

  • Formatting Rule: When drafting code or schemas, use modular "Lego-brick" architecture so new features (like a Donation button) can be added later without a rebuild.

How to use this memory:

  1. For Claude/ChatGPT: Add this to your "Custom Instructions" or a project-specific markdown file.

  2. For OpenBrain/Supabase: Use a tool like capture_thought to store this under the category project_context.

  3. For Cursor/Windsurf: Place this in a file named .ai_memory or PROJECT_CONTEXT.md in your root directory.

Consultant's Note: This memory ensures your AI never suggests a "manual-only" solution again and remembers that Excel integration is the heart of your efficiency upgrade.