An AI automation prioritization framework for investment firms
Direct answer: score every candidate workflow on value, feasibility, and risk. Multiply value by feasibility, subtract risk, and run the highest scoring workflow first as a contained pilot with a named reviewer. Most firms get this wrong by chasing the most exciting use case instead of the best balanced one.
The three axes
| Axis | What you are measuring | Score 1 to 5 |
|---|---|---|
| Value | Hours saved per month, revenue or fundraising impact, partner time freed, error rate reduced | 5 = clearly material to the P&L |
| Feasibility | Inputs already exist in structured form, rules can be written down, a named reviewer exists, system access is available | 5 = could ship a pilot this quarter |
| Risk | Regulatory exposure, fiduciary duty, client visibility, reversibility of an error | 5 = an error is hard to recover |
Score = (Value x Feasibility) - Risk. Anything below 8 is not your first project. Anything above 15 is usually a fantasy: re-score it honestly.
A worked example
DDQ response drafting. Value 5 (IR team is the bottleneck on every capital raise). Feasibility 4 (prior DDQs, ADV, PPM, side letters all exist; IR already reviews every answer). Risk 2 (every answer is reviewed and signed off before anything goes to an allocator). Score = (5 x 4) - 2 = 18. Strong first candidate.
Compare with autonomous deal screening that emails partners directly. Value 4. Feasibility 2. Risk 5. Score = (4 x 2) - 5 = 3. Wrong first project, even though it sounds more impressive.
How to actually run the exercise
List 8 to 12 candidate workflows from IR, deal, research, ops, and compliance. Score each one with the people who actually run them, not with the executive sponsor. Rank by score. Pick the top one as the pilot, the next two as the roadmap, and shelve the rest until you have shipped one.
What to commit to before you start
A named reviewer in every approval lane. A single source of truth for the inputs. A defined success metric (for example, first reviewable draft in under one hour, with every claim cited to a versioned source). A kill criterion if the metric is missed twice in a row.
Next step
Want help running this on your shortlist?
Read the companion guide on how to identify candidates, or book a workflow audit and we will score the top candidates with your team.