Business of AI · Gallery

AI Initiative Portfolio Dashboard

SIMULATEDVerified Jul 2, 2026

A book of twelve AI initiatives, governed like capital. Each carries a risk-adjusted ROI and an explicit call — because a portfolio where nothing is ever killed isn't governed, it's unattended.

Same instrument · three industries — pick a use-case to reconfigure the run

Expected annual value

$26.7M

Sum of unadjusted upside

Run-rate spend

$10.4M

Annualized

Risk-adjusted value

$7.4M

Value × P(success) − spend

Recommend to kill

2/12

Negative risk-adjusted ROI

Value × risk · bubble = run-rate spend

high valuelow valuehigh risk →

Scale Hold Kill

KYC document intelligence

Finserv · pilot

Kill

Expected value

$1.4M

P(success)

30%

Run-rate spend

$1.1M

Risk score

0.75

How this number is computed

Risk-adjusted ROI = expected value ($1.4M) × stage probability (30% for pilot) − run-rate spend ($1.1M) = -$0.7M/yr.

Call: negative risk-adjusted ROI → kill.

2 to cut this quarter

2 of 12 carry a negative risk-adjusted ROI. Keeping them funds optionality theatre — the capital they hold is the capital the Scale column needs.

Steering-committee takeaway: A portfolio where nothing gets killed isn't governed — it's unattended. Two of these twelve should die this quarter.

How this is built & assumptions

Stage probabilities (editable defaults): discovery 15% · pilot 30% · scaling 60% · production 85%.

Risk-adjusted ROI = expected annual value × stage probability − run-rate cost. Kill if < 0; scale if proven-stage and ≥ 1.5× spend with risk < 0.6; else hold. Plan variance flagged beyond ±10%.

Stack: Next.js (static) + shared design system; deterministic client-side math over authored, anonymized finserv + telecom initiatives.

Limitations: probabilities are stage-based defaults, not per-initiative Bayesian estimates; value figures are illustrative. The instrument frames the allocation decision — it doesn't replace the finance model behind it.