What we were testing ?
The initial experiment started from AI-augmented portfolio management: could AI help organisations leverage their portfolio management and improve future prioritization?
The discussion exposed a more fundamental problem: most organisations already have portfolio information where they track statuses, budgets, milestones, risks and KPIs. Yet these elements do not necessarily explain why an investment remains justified.
The experiment therefore shifted from portfolio reporting toward portfolio decision traceability.
What we observed ?
A project can remain green while the original value hypothesis weakens and a business case can continue to exist after its assumptions have changed.
More importantly, a portfolio can contain large amounts of data without preserving the reasoning that connected that data to an investment decision.
We therefore started separating facts, assumptions, value claims, evidence and decisions with a clear objective in mind:
- What value was expected?
- What evidence supported that expectation?
- Which assumptions were required?
- What decision was made from that evidence?
- What has changed since then?
- Does the current evidence still support the investment?
What changed in the design ?
The concept moved away from another portfolio dashboard to focus instead on the evidence chain behind portfolio decisions.
A simplified representation is:
Strategic intent
↓
Value claim
↓
Supporting evidence
↓
Assumptions
↓
Decision
↓
Observed evolution
↓
Reassessment
AI becomes useful here because it can help identify broken links:
- unsupported claims;
- stale assumptions;
- inconsistent evidence;
- decisions whose original justification is no longer visible;
- initiatives whose value narrative has changed without explicit reconsideration.
The system does not need to make the portfolio decision; it can make the reasoning behind that decision easier to inspect.
What we retain ?
Portfolio data and portfolio reasoning are different assets : Better reporting can improve visibility without improving decision quality, while value traceability requires preserving not only the outcome of a decision, but enough of its reasoning to challenge it later.
Takeaway
The useful AI opportunity shouldn't be an automated portfolio selection : It should instead focus on making the chain between value, evidence and investment decisions explicit enough to be reviewed.