A promising demo that fails in practice
The AI or automation works in a demonstration but does not produce the expected result in the real workflow.
AI and automation diagnostic
Before You Spend More on AI or Automation, Figure Out What Is Actually Failing.
This is for teams and decision-makers facing an AI adoption, agent, automation, implementation, workflow, product, or strategy problem where it is unclear whether the failure is technical, operational, commercial, or organizational.
In a focused 60 to 90 minute session, I will help you separate the competing explanations, identify what does not fit, and determine what deserves attention next.
$495. One problem. One focused session. One concise written assessment.
Not sure whether your problem fits? Tell me about it first. I’ll tell you whether I think this diagnostic is an appropriate next step before you spend anything.
AI and automation problems this is designed to clarify
This is for you if the technology is only one part of the problem.
Adoption that keeps stalling
Competent people are not using the system consistently, and nobody agrees whether the problem is training, incentives, usability, trust, or workflow design.
Automation that moves the bottleneck
One step became faster, but the overall process did not improve or created a new failure somewhere else.
A failure with several possible owners
It is unclear whether the issue belongs to the model, vendor, data, product, implementation, operating process, or commercial strategy.
Different teams, different explanations
Technical, operational, product, and leadership teams are describing different versions of the same failure.
These are examples of AI and automation implementation, workflow, adoption, product, and strategy problems. This $495 Diagnostic is separate from the $7,500 Difficult Problems Evaluation Sprint.
A separate commercial path
Need rigorous AI reasoning evaluation?
The separate $7,500 Difficult Problems Evaluation Sprint tests one defined AI system, agent, workflow, or model implementation against purpose-built difficult reasoning cases. It is not part of this $495 Diagnostic.
Outside judgment and synthesis
What I will do with you
You bring me the intended outcome, current workflow, system behavior, implementation history, operating constraints, and competing explanations. I will ask questions, test assumptions, look for contradictions and hidden dependencies, and try to identify whether the problem is technical, operational, commercial, organizational, or an interaction among them.
After the session, I will send you a concise written assessment covering:
- 1What I think is most likely going wrong
- 2The primary bottleneck or failure point
- 3What does not fit or still needs to be verified
- 4The risks that deserve attention first
- 5The next actions I would prioritize
- 6Whether the problem appears to need specialist or larger implementation work
A guided pre-session intake
You do not need to build a perfect case file.
Before the 60 to 90 minute session, you will complete a structured intake and may provide up to five key supporting items. The purpose is to help me understand the background before we meet so the live session can focus on the actual problem rather than basic fact gathering.
Don’t worry about knowing exactly what to send. Start with the facts you have. I’ll guide you.
You do not need to diagnose your own problem before meeting me. The intake is designed to help me understand the situation even if you are not sure what matters yet.
The intended outcome
A concise description of what the AI or automation was supposed to improve and how success would be recognized.
The current workflow and system
A workflow description, system overview, process map, or other concise explanation of how the work happens today.
Evidence of the gap
Relevant operating measures, output examples, adoption data, quality findings, user feedback, or another bounded view of what is not working.
The implementation timeline
A short chronology of what was selected, built, changed, launched, and when the problem became apparent.
Competing explanations or proposed fixes
Vendor recommendations, technical assessments, product proposals, internal memos, or other material showing what different people think should happen.
If you do not have all five, that is fine. Send what you have. Do not delay trying to create perfect information.
Do not send passwords, API keys, access tokens, customer records, confidential legal files, medical records, protected health information, card data, or other highly sensitive material through the initial intake.
If material beyond this bounded review is needed, I may ask for one or two additional items later or recommend a larger engagement.
A concrete written deliverable
What your written assessment looks like
What I think is actually happening
The explanation that best fits the information available.
Why I think that
The facts, contradictions, dependencies, or assumptions driving that conclusion.
What still doesn't fit
Important unanswered questions and things that still need verification.
What I would do next
The actions I would prioritize, and what I would deliberately not spend time or money on yet.
Why Sonny
Why bring the problem to me?
My working life has repeatedly put me in situations where the important question was not “Who is the specialist?” but “What is actually happening?”
Emergency medicine trained me to make decisions with incomplete information, competing explanations, time pressure, and real consequences. I now apply the same discipline to difficult problems across business, commercial real estate, organizations, AI, workflows, and systems.
Review the method first
Want to see how I think before hiring me?
Review my public Difficult Problems work to see how I examine competing explanations, test assumptions, identify what does not fit, and decide what deserves attention next before you purchase anything.
Rapid Systems Integrity Diagnostic
$495
One-time fixed fee
- 1Structured pre-session intake and review of your key facts and supporting material.
- 260 to 90 minute diagnostic session
- 3Concise written assessment
- 4Primary bottleneck / failure point
- 5Important unresolved questions
- 6Prioritized next actions
Not sure whether your problem fits? Tell me about it first. I’ll tell you whether I think this diagnostic is an appropriate next step before you spend anything.
Larger implementation or advisory work is separate. You are not committing to anything beyond this diagnostic.
A finished first engagement
What happens after you buy
- 1You complete the secure $495 checkout.
- 2You receive the structured intake and scheduling instructions.
- 3You give me a concise description of the problem and the information needed for the session.
- 4We spend 60 to 90 minutes working through it together.
- 5I send you the written assessment and prioritized next steps.
Payment is processed securely by Stripe.
A useful boundary
When this is not the right first step
- 1You already know exactly what the technical problem is and simply need an engineer, vendor, or implementation team to execute the solution.
- 2You specifically need a rigorous evaluation of an AI system, agent, workflow, or model implementation against purpose-built difficult reasoning cases.
- 3You need emergency security response, legal advice, regulatory advice, medical advice, or another regulated professional service.
- 4You want a guaranteed technical, operational, commercial, organizational, or investment outcome.
If you are unsure, discuss the problem first.
This is a fixed-scope diagnostic and strategic advisory engagement. It is not a model benchmark, security audit, technical certification, legal opinion, or guarantee of an AI or automation outcome. The separate $7,500 Difficult Problems Evaluation Sprint is a different engagement.
Do not submit passwords, API keys, access tokens, customer records, confidential legal files, medical records, protected health information, card data, or highly sensitive material through the initial purchase or intake flow.
If the cost of scaling the wrong AI or automation solution is more than $495, this is a sensible place to start.
Bring me one difficult AI or automation problem. I will help you determine what deserves attention next.
Not sure whether your problem fits? Tell me about it first. I’ll tell you whether I think this diagnostic is an appropriate next step before you spend anything.