Diligence · Operating decision guide
Which portfolio-company AI initiative gets the next dollar?
Otomat editorial · Revised
Remove proposals without an owner, rights or a testable baseline before comparing upside. Then allocate scarce management capacity to the smallest next test that can change a decision.
Use gates before a weighted score
A large estimated benefit should not offset missing source-data rights or the absence of an operating owner. Mark those proposals as held, with a specific action to unblock them. For eligible proposals, compare the business constraint, current cost of delay, available people, test cost and downside. Keep scenario value separate from observed benefits when discussing priority.
Compare the next commitment
One proposal may ask for a bounded assessment while another asks for a multi-system rollout. Put the next decision gate and required spend on the same page before ranking them. A short test that can disprove a fragile assumption may be valuable even if it does not create a reusable platform. Avoid scoring precision that the underlying evidence cannot support.
| Check | Evidence to bring | Decision or stop condition |
|---|---|---|
| Eligibility | Local owner, lawful data access and observable baseline | Hold a missing prerequisite regardless of the upside estimate. |
| Next commitment | Specific test, full cost and information it will produce | Prefer a decision-changing test over an unbounded implementation. |
| Shared capacity | Named reviewers and overlapping demands | Sequence initiatives that compete for the same scarce people. |
| Reassessment | Observed result and reason to continue or stop | Re-rank on evidence rather than the original presentation score. |
Account for shared bottlenecks
List the finance, security, integration and frontline reviewers needed across proposals. Two attractive projects can be poor concurrent choices when they consume the same scarce team. Ask whether a common prerequisite, such as source cleanup or identity repair, should come first. Shared preparation is not permission to combine confidential data from different companies.
Revisit the order when evidence changes
Record why a proposal moved ahead and what would reverse the decision. When the test concludes, update the gate instead of defending the original score. Stop or narrow initiatives that cannot secure the dependency they need. Compare accepted outcomes at each company; do not add hypothetical portfolio benefits as though every proposal will be funded and succeed.
Questions before the decision
Should the biggest estimated ROI win?
No. Estimates with different evidence quality, dependencies and downside are not comparable. Clear the prerequisites and compare the next commitment before debating scale.
Can a sponsor impose one score across companies?
A shared decision format can help. Local baselines, priorities, rights and management capacity still determine whether a project is viable.
Take the evidence sheet to the meeting
Download the blank CSV, assign an owner to each check and record the evidence. Keep completed sheets in your own approved system.
Download the evidence sheet (CSV)Use the working tools
Blank templates are ungated. Adapt them to the workflow; inputs are not submitted.
Sources and limits
These are Otomat’s proposed operating methods and original decision artifacts. External references inform risk management; they do not certify this method or establish client results. Linked sources were selected for this review; their live revision dates may change.
- NIST AI RMF: governance, measurement and risk management context
- US GAO: AI accountability framework and evaluation context
For anonymous documentation and proposal patterns, see delivery evidence and its boundaries.
Related decisions
Bring a workflow to the discussion
Share the business goal and constraints through the mandate process. Do not send sensitive production data in the initial brief.
Book a Return Review