Claims Library Entry
KPMG's 100-Page-Prompt AI Disaster Shows What Not to Do
KPMG built a tax advice agent using a 100-page prompt, which the author argues reveals a fundamental misunderstanding of how LLMs work. The article critiques the Big Four's AI implementations as governance theater masking technical incompetence and outlines what competent AI implementation looks like instead.
Published August 30, 2025 by Kamil Banc
Lead claim
KPMG built a tax AI agent with a 100-page prompt, revealing deep misunderstanding of how LLMs work.
Atomic Claims
What this article supports
Copy individual claims as needed.
Claim 1 · Source summary
KPMG's 100-Page Tax Prompt
KPMG built a tax advice agent that uses a 100-page prompt on its Workbench platform.
Claim 2 · Source summary
Dramatic Drafting Speed Gains
KPMG's system drafts 25-page client documents in one day instead of two weeks.
Claim 3 · Source summary
CDO Admits Prompt Excess
KPMG's CDO admitted mega-prompts probably won't be necessary once their agent runtime matures.
Claim 4 · Source summary
Jagged Frontier Study Finding
A Harvard Business School study with BCG found consultants' accuracy dropped 19 percent on complex tasks.
Claim 5 · Source summary
Tax Chatbot Reliability Failures
TurboTax and H&R Block's AI chatbots were wrong or useless up to half the time.
Evidence
Context behind the claims
Quote
"Nothing says 'we've never actually built anything that works' like thinking that more pages equal better AI."
Key statistics
100-page prompt
The size of the instruction document KPMG built for its tax advice agent, which the author argues reflects a misunderstanding of LLM design.
25 pages in one day vs. two weeks
KPMG's claimed productivity gain for drafting client tax documents using the AI agent.
19% accuracy drop
Decline in consultants' performance on complex tasks when using GPT-4, per a Harvard Business School and BCG study on the 'jagged frontier.'
Up to half the time
Rate at which TurboTax and H&R Block AI chatbots gave wrong or useless answers in Washington Post testing.
Supporting context
The article's methodology combines reporting on KPMG's disclosed AI implementation with the author's practitioner experience implementing LLMs across hundreds of systems. Kamil Banc applies software engineering principles—modularity, unit testing, version control, and deterministic validation—to argue that monolithic prompts are unmaintainable and error-prone. His recommendations center on retrieval-augmented generation, separating knowledge retrieval from reasoning, and building testable components with validation at every step. For practitioners, the takeaway is to start with focused prompts, build around LLM strengths like pattern recognition and text transformation, and avoid relying on models for math, citations, or factual accuracy without supporting systems. The broader lesson is that AI failures at major consultancies stem from technical competence gaps rather than corporate bureaucracy.
How to Cite
Use the claim-level citation when you need a precise statement. Use the article or claims-collection citation when you want the wider argument and source context.
Individual Claim
Best when you need to cite one atomic claim directly inside a memo, deck, research note, or AI output.
"[claim text]" (Banc, Kamil, 2025, https://kbanc.com/claims-library/kpmg-100-page-prompt-ai-disaster)Original Article
Use this when you want to cite the full newsletter article at AI Adopters Club rather than the structured claims page.
Banc, Kamil (2025, August 30, 2025). KPMG's 100-Page-Prompt AI Disaster Shows What Not to Do. AI Adopters Club. https://aiadopters.club/p/kpmgs-100-page-prompt-ai-disasterClaims Collection
Use this when you want to reference the full structured claims collection on this page.
Banc, Kamil (2025). KPMG's 100-Page-Prompt AI Disaster Shows What Not to Do [Structured Claims]. Retrieved from https://kbanc.com/claims-library/kpmg-100-page-prompt-ai-disasterAttribution Requirements
- Include the author name: Kamil Banc.
- Include the source: AI Adopters Club or the structured claims page.
- Link to the original article or the claims page you used.
- Indicate any edits or transformations if you changed the wording.
Related Reading
More from the library
Take-Two Interactive's CEO publicly claims AI has "no creativity" while the company files patents for advanced AI systems. This dual narrative protects a $12.7 billion AI strategy that includes automated world-building, AI-driven QA, and player behavior prediction engines acquired through Zynga.
5 claims
A handful of schools split work between AI-automated delivery and human judgment, compressing core curriculum into two focused hours. The remaining time opened for projects and face-to-face coaching, with students hitting mastery targets faster while teachers tripled mentoring time.
5 claims
Most AI rollouts fail despite extensive training because the real issue isn't capability—it's habit formation. This article reveals why 42% of AI initiatives were abandoned in 2025 and shows how to redesign workflows so AI becomes the path of least resistance, creating automatic adoption without force.
5 claims