Committee Board Agentic Application

I led the end-to-end product design of a platform that transforms how committees across DGE review proposals, deliberate, and reach decisions — representing ICD as the design partner for UnifyApps, working directly alongside Frog Design Agency and Nagarro Design. The platform replaces fragmented manual workflows with structured AI intelligence at every stage, from proposal ingestion through to resolution, serving every committee, department, and entity under DGE.

ADEO managed over 1,000 high-impact submissions monthly across its committees and the Executive Council. Proposals came through ECAS — digitally submitted, mostly in Arabic, as PDFs and presentations. But beyond submission, the process was entirely manual. Coordinators reviewed documents by hand, analysts drafted summaries independently, and approvals moved through multiple levels before a proposal ever reached the committee room — a process that took up to 15 working days. Four problems were structural:

Research was a continuous process throughout the engagement — not a single phase. I gathered requirements directly from the client, worked through multiple versions and design iterations based on stakeholder feedback, and had direct conversations with various departments to understand their specific needs. The product team was involved at every stage, and every major decision was validated with the client before moving forward. Three user types emerged as the core of the design problem:

The platform automatically analyses every proposal before the session begins — generating a summary, strategic alignment score, ROI estimate, feasibility rating, and risk assessment — so members receive a complete briefing without any manual preparation. It also connects to a 5-year archive of historical committee decisions and policies, making them searchable in plain language during a live session. Everything surfaces directly in the view the member is already working in.

Session Agenda

One view. Every session. Every decision that matters.

Members open the platform and see only their committees — the active session at the top, upcoming sessions below. Every proposal in the session is listed with its type, presenter, and ADEO representative, split clearly into For Information and Decision Making. I kept each member's view scoped to their own committees only. It made the moment of arrival simple and focused. The cost was that members with responsibilities across multiple committees had to access a separate view — a trade-off worth making because that need was the exception, not the rule.

AI Analysis & Session Briefing

Structured navigation. Intelligent briefing. One surface.

Selecting any agenda item opens a structured view — AI-generated summary, analysis, team opinion, and draft resolution — all prepared from the source documents before any member opens it. I placed the AI content at the top, not buried behind the raw documents. The cost was that trust had to be earned at the first moment of contact, with no separate explanation to fall back on.

AI Assistant & Document Intelligence

Five years of institutional knowledge. Queryable in plain language.

The platform connects to a 5-year archive of 3,000–4,000 historical committee records and policies, searchable through a conversational assistant within the session view. A member can ask whether a similar proposal was previously approved and get a sourced answer without leaving the meeting. On tablet, the assistant and proposal content compete for space — I gave the proposal priority and kept the assistant collapsed by default.

Voting, Survey & Async Participation

Every voice captured. Every decision accountable.

Members vote through five explicit outcomes — Approve, Approve with Comments, Not Approved, Return for Committee, or Abstain — each requiring a documented reason before submission. Members who miss the session can vote and submit feedback across four dimensions after the fact. I made voting deliberately slow. Every option forces specificity — no rejection without a reason, no approval without attribution. The cost was friction. In government decision-making, that friction is the point.

Replacing a 15-working-day manual preparation cycle with automated AI analysis changed how fast committees moved, how confidently they decided, and how completely every outcome was recorded. The platform did not speed up the existing process — it replaced the parts that should never have been manual in the first place.

15 days → minutes

The entire pre-committee preparation chain replaced by an AI-generated briefing that arrives before the session begins.

3,000–4,000 docs

Five years of institutional knowledge — previously inaccessible during deliberation — now queryable in plain language mid-session.

1,000+ / month

High-impact government proposals managed monthly — each one analysed, deliberated, voted on, and documented within the same platform.

0 post-session drafting

Minutes, action items, and resolution drafts generated live during every session — complete before the room clears.

This project taught me that AI alone doesn't create value—trust does. The platform supports decisions that influence public policy and resource allocation across Abu Dhabi, so every AI-generated insight had to be transparent, explainable, and easy to verify. Every design decision was made with one goal: helping people trust the system at the moment they use it.