Four tools
Use them before we ever speak. No form, no email required, no commitment.
Program diagnostic
Select the scenario that best describes your program, or describe it in your own words. Ricardo's AI will show you how he would approach it based on real engagements.
How this works →
A language model working from a brief I wrote out of my own engagements. Worth knowing what is behind it before you weigh the answer.
- What it knows. My actual record at FICO, Honeywell, S&P Global, Google Cloud and Mayo Clinic, plus the sequence I run in a recovery: executive status live on day one, governance reset and decision rights by week two, risk register with named owners by week four, operating cadence rebuilt by week eight.
- What it does with what you write. It reads your situation and answers the way I would: whether the failure looks structural or executional, what I would stabilise first, and what the first thirty days would look like. One sentence gets you two paragraphs; real detail gets you a real answer.
- What it is not. A diagnosis of your program. It has never seen your plan, your team or your numbers. It is matching your description against my patterns. Treat it as the first five minutes of a conversation, not the conclusion of one.
What you write reaches my inbox, which is how I see whether the tool is any good. Nothing is published or shared with anyone. Do not write anything you would not send me in an email.
Status report translator
Paste the status report you received this week. You get what it is actually saying, and three questions to ask at the next steering committee.
AI pilot exposure calculator
Your pilot has been running longer than planned and nobody has made a production decision. This puts a number on what that costs, using published abandonment rates rather than my guesses.
Where every number comes from →
You supply the money. The model supplies only published percentages, and here is each one with its source.
- You give it the spend. Pilot cost, headcount and monthly rate are yours. I did not benchmark them, because the public figures for what an AI pilot costs range from fifteen thousand to five hundred thousand dollars, which is a price list rather than a benchmark. You already know your number better than any of them.
- The model gives the odds. Gartner: at least 30% of generative AI projects are abandoned after proof of concept, and through 2026, 60% of AI projects unsupported by AI-ready data will be abandoned. MIT State of AI in Business 2025: of firms that evaluated an enterprise system, 60% evaluated, 20% piloted, 5% went live, and 95% of pilots produced no measurable P&L impact.
- Which rate applies to you. A named production owner and under twelve months puts you on the 30% figure. No named owner, or past twelve months, moves you to 60%. Both, past eighteen months, applies the 95% no-impact finding. The thresholds are a judgement I am making in the open, not something Gartner published.
- What it deliberately does not do. It does not estimate the value you would have captured, or price your reputational risk, or tell you the pilot is doomed. It multiplies money you have already spent by a failure rate somebody credible measured. That is a floor, not a forecast.
This one is different from the others: it runs entirely in your browser and nothing leaves your machine, not even to me. If the number is uncomfortable, the useful response is not to argue with the percentage. It is to go and find out whether your pilot has a named production owner.
Cost of delay calculator
Enter your program details. Numbers update in real time based on industry benchmarks for overhead burn and opportunity cost.
How this is calculated →
Two inputs of yours, one arithmetic operation, and one disclosed rate. No industry multipliers, because I could not source any.
- Weekly burn. Budget divided by duration. That is what the program spends every week it stays open, and it is arithmetic, not an estimate.
- The benefit that does not arrive. The annual benefit you expected, divided by 52. Every week the program does not finish is a week of that benefit you do not collect. This is the whole idea behind cost of delay, and the number is yours rather than a factor I invented.
- Weekly cost of delay. The two added together. Multiply by weeks of slip for the accumulated figure, or by 30, 60 and 90 days for the projection.
- What the intervention costs. A twelve week recovery at eight thousand dollars a week, which is the middle of my C2C range. Rather than showing you a return multiple, the tool converts that cost into weeks of your own delay. If it comes out under a week, that is not me being persuasive. It is your own numbers.
Earlier versions of this tool multiplied burn by an industry factor and reported a return multiple. It produced results above 200x, which is the kind of number that makes a CFO stop reading. The factors were mine and I could not point to a source, so they are gone. Everything on screen now is either your input or a rate I have told you.
The first 30 days of every recovery follow the same structure.
