← All field guides

A BlueAlly Field Guide

What to build
first.

Most AI portfolios are ranked by who asked loudest, not by what pays. This is the cure. Score every idea by value, readiness, and risk — then plot it, and let the picture decide.

Conquer Complexity

READINESS → VALUE → Champions Rank it honestly. The corner that earns the money is the one to start.

What's inside

01  The premise

Almost everyone is doing AI. Almost no one is getting paid for it.

Eighty-eight percent of companies now use AI in at least one part of the business. Only about six percent can point to real money on the bottom line.1 The gap is not the models. The gap is the list.

Most AI roadmaps are ranked by who asked loudest, which executive sponsored it, or which vendor demoed best. The result is a pile of pilots that never scale, and a budget spent on the wrong three things. Two out of three companies have not moved a single use case past the pilot stage.1

There is a better way, and it is not complicated. Score each idea on the few things that actually decide its fate. Plot the scores. Then start where the value and the readiness are both high — and leave the rest for when they are ready. McKinsey calls it a value-versus-feasibility matrix. We just call it telling the truth.2

88%
of organizations use AI in at least one business function — up from 78% a year earlier.1
~6%
are "AI high performers" — the few attributing more than 5% of profit to AI.1
have not yet scaled a single use case beyond the pilot. The list, not the model, is the bottleneck.1
A backlog ranked by politics is a budget spent on the wrong three things.

02  The method

Honest ranking, in three moves.

You do not need a data-science team to prioritize well. You need a shared scale, applied to every idea, in the open. Three moves do it.

In plain English

Three words to carry the whole tool

Value
What the idea is worth in a year — cash saved, revenue earned, risk avoided. Measured as expected value over the friction it removes, not a guess. One number, in dollars, that you can defend.
Readiness (feasibility)
How ready you are to ship it. Scored on four weighted pillars — Organization 35%, Data 30%, Governance 20%, Technical 15% — each rated 1 to 10. Six is the line: above six, deploy with confidence; below six, build readiness first.
Risk
What could bite — a regulator, a wrong answer, a security hole. Risk does not kill an idea. It sets the guardrails and drags down the priority you can count on.

Move one — discover. Move two — score. Move three — prioritize.

First you surface the opportunities: where does work stall, and what kind of AI would fix it. Then you score each one on value, readiness, and risk — the same scale for all of them. Then you plot them on one field and read the corners. The tool below does all three. The examples are illustrative, drawn from a national private-equity firm's portfolio — real shape, neutral names.

What sets the score apart — the priority model and how readiness is weighted THE PRIORITY MODEL Priority = Value × Readiness × Confidence − Risk Drag Value = Expected Value ÷ friction cost Return measured against the friction it removes, not a guess. Confidence discounts a shaky case; risk drags the priority you can count on. READINESS, WEIGHTED Organization 35 Data 30 Governance 20 Technical 15 Weight of the readiness score (%) 6 The 6.0 threshold Above six, deploy with confidence. Below six, invest in readiness first.
Fig. 1 — The model behind the score. Priority weighs value by readiness and confidence, then subtracts risk drag. Readiness itself is four weighted pillars — Organization 35, Data 30, Governance 20, Technical 15 — and six is the line between piloting and producing.
From a messy backlog to a clear plan — three moves 1 · DISCOVER Surface the work Friction points AI building blocks Business drivers 2 · SCORE One honest scale Value Readiness Risk 3 · PRIORITIZE Build the green corner first The same scale on every idea is what makes the picture trustworthy — and the decision defensible.
Fig. 2 — Discover, score, prioritize. Surface the work, score it on one scale, then plot it. The corner where value and readiness are both high is where the money is — and where you start.

03  The working tool

Rank it honestly. Click through it yourself.

A live, seeded portfolio for an example firm. Walk the three steps. Re-score a use case and watch its place on the matrix move. Filter the field. The math is deterministic — no black box, no guesswork.

AI Opportunity Portfolio Example: a national private-equity firm · 10 seeded use cases
Portfolio value / year$0M

Before you score, you name the work. Every use case is described in four plain lenses. These are the questions a good discovery asks — and the dimensions the score is built from.

The friction

Where does work stall today? Name the pain before the fix. Most value hides in the boring places.

Delays & handoffsRework & errorsData silosRepetitive workKnowledge gapsSeasonal peaks

The building block

What kind of AI does the job? A handful of patterns cover most enterprise work. Pick the ones that fit.

Content creationRetrieval (RAG)ClassificationExtractionForecastingWorkflow automation

The business driver

Why does the business care? Tie every idea to one of four outcomes a CFO already tracks.

Grow revenueCut costFree cash flowReduce risk

The risk weight

What raises the stakes? Some domains carry more. We weight them, so a regulated idea is judged on a fair scale.

Regulatory ×1.5Cybersecurity ×1.4Reputational ×1.3Financial ×1.2Operational ×1.0
So what: the four lenses are not paperwork. They are exactly the inputs the score needs — friction and building block point to value, the driver names the dollars, and the risk weight keeps the comparison honest. Move to Score to see them turned into numbers.

Dollar figures and scores are illustrative, built to show the method. The model is real and shown in full: Priority = Value × Readiness × Confidence − Risk Drag, where Value is expected value over the friction it removes. Readiness is four weighted pillars — Organization 35%, Data 30%, Governance 20%, Technical 15% — and six is the line between piloting and producing. This live demo computes a transparent, deterministic blend of those same factors.

04  Reading the picture

Four corners. Four different decisions.

A 2×2 is only useful if each corner tells you to do something different. This one does. The line that splits the field is "ready enough to ship in months." The line that crosses it is "worth enough to fight for."

1
Champions — high value, high readiness. Deploy now.

The money is here and so is the runway. Fund these first, this quarter. If a green-ringed bet sits here, it leads the roadmap.

2
Strategic — high value, readiness gap. Plan a sprint.

Worth a lot, not ready yet. Do not start building. Sponsor a 90-day sprint to close the gap — the data, the workflow — then promote them.

3
Quick Wins — modest value, high readiness. Ship fast.

Small but shippable. Use them to build the muscle and prove the loop. A run of quick wins funds the bigger bets.

4
Foundation — lower value, lower readiness. Build first.

Not yet, and that is fine. Note them. Revisit when the data is cleaner or the capability arrives. Saying "later" is a decision too.

The matrix does not make the decision. It makes the decision obvious.

05  The close

A list became a plan.

You started with a backlog ranked by politics. You scored it on three honest dimensions. You plotted it. Now the next three things to build are not a matter of opinion — they are the green corner, and you can defend every one of them to the board.

That is the whole move. The scale is simple. The discipline is rare. The hard part was never finding AI ideas — everyone has a hundred. The hard part is the judgment to rank them honestly, fund the few that pay, and have the nerve to say "later" to the rest. That judgment is what BlueAlly brings to the table.

← The AI Decision Frameworks

06  Sources

Where this comes from

The adoption figures are from McKinsey's most recent global AI survey. The value-versus-feasibility method is theirs as well — we have built a working tool on top of it. Dollar figures inside the tool are illustrative; the scoring weights and the method are not.

  1. McKinsey & Company, "The state of AI in 2025: Agents, innovation, and transformation" (88% AI adoption; ~6% high performers; the scaling gap). mckinsey.com/.../the-state-of-ai
  2. McKinsey & Company, "From promising to productive: Real results from gen AI in services" (value-versus-feasibility prioritization). mckinsey.com/.../from-promising-to-productive
  3. McKinsey & Company, "A data leader's operating guide to scaling gen AI" (priority-domains strategy; value, feasibility, strategic fit). mckinsey.com/.../scaling-gen-ai
  4. Anthropic, "Claude — Models overview" (Claude Opus 4.8, 1M-token context; the deterministic-where-it-counts pattern). platform.claude.com/docs/en/about-claude/models/overview