K
Enterprise Operations2025Portfolio Management
This project is delivered under NDA. Client and product identities are withheld; screens are access-gated and shown to prospective clients on request.

AI Portfolio Command Center

An AI-driven decision platform that helps portfolio leadership detect delivery and commercial risk early, decide faster, protect margin and improve fulfilment — built explicitly as a decision system rather than a reporting layer.

Product Design Lead & Frontend Engineer
2025
Enterprise Technology Services
Client:Portfolio Leadership
ReactTypeScriptPythonLLM integrationRecommendation logicDashboard design
Problem Statement

A decision system for portfolio leadership — not another reporting dashboard.

Our Approach

The platform models demand, pipeline, resourcing, budget and approvals as one connected chain, so recommendations reason across the whole picture instead of one isolated slice.

AI acts as a prioritisation and early-warning layer — surfacing emerging risk and proposing options, while approval authority stays entirely with the people who hold it.

Every view was designed to answer a specific recurring leadership question. Metrics that did not lead to an action were deliberately excluded.

Challenge

Portfolio leadership had reporting in abundance and decision support in short supply. Dashboards showed what had already happened, by which point margin erosion or a fulfilment gap was expensive to correct.

Risk signals existed but were scattered across demand, staffing, budget and delivery systems — and nobody could see the chain end to end in time to act.

The brief was explicit and unusually disciplined: this is not a reporting dashboard. If a screen did not change a decision, it did not belong.

Solution

An AI-driven decision platform that helps portfolio leadership detect delivery and commercial risk early, decide faster, protect margin and improve fulfilment — built explicitly as a decision system rather than a reporting layer.

Outcome

For leadership teams who have plenty of reporting and still find out about problems late — the fix is usually a connected model and decision-shaped views, not another chart.

The constraint that shaped it

If it doesn't change a decision, it doesn't ship

The hardest discipline was refusing the metrics stakeholders asked for out of habit. Leadership dashboards fail by showing everything and prioritising nothing — so every screen had to justify itself against a decision someone actually makes.

· Decisions that mattered

Recommend, never auto-assign

Portfolio decisions carry commercial and human consequences. The system's job is to make the right option obvious and the risk visible, not to make the call.

One connected model, not four dashboards

Most of the value came from modelling the full chain. Recommendations over a single disconnected slice would have been shallow enough to ignore.

· What I owned
  • Led product definition and the decision-system framing
  • Designed the leadership experience and information architecture
  • Built front-end implementation and integration
  • Designed how AI recommendations surface so they inform rather than pre-empt decisions
· Design Process

How We Solved the Problem

01
01

Connected portfolio model

Demand, pipeline, resource availability, budget and approvals modelled together, so commitment and capacity are visible against each other rather than in separate systems.

    02
    02

    Risk detection layer

    Signals across the chain are evaluated for emerging fulfilment and margin risk, surfaced early enough that the decision still has options.

      03
      03

      Recommendation engine

      Proposes fits and interventions with the reasoning attached, presented as options for a decision-maker rather than automated actions.

        04
        04

        Decision-first dashboards

        Each leadership view answers one recurring question and prioritises exceptions over totals.