Your AI sounds confident. Should you believe it?

When you ask your AI agent “why did you make that decision?”, it always has an answer. That answer sounds reasonable. Sometimes insightful.

But how much of it is real?

We built a formal framework to measure it. The finding: AI can produce self-explanations 10 layers deep. Only 2.5 are externally verifiable. The other 7.5 levels are a confabulation zone - structurally indistinguishable from genuine reasoning.

This 90-minute workshop gives your team the tools to know where to trust AI and where to verify.

What Your Team Gets

Live Demo

Two AI instances, same model, different contexts. Your team designs the prompts. Watch in real-time as self-referential depth degrades - and learn to spot exactly where.

The Three Numbers

Every AI deployment needs these:

• r_represent: how deep AI can produce explanations (spoiler: very deep)
• d: how deep it can *internally verify those explanations
• r_valid: how deep those explanations actually hold up

Practical Guidelines

Five rules your team applies Monday morning:

1. Depth-aware prompting (ask for reasoning at depth 1-2 only)
2. External verification architecture for anything deeper
3. Confabulation zone alerts in your AI pipelines
4. The vacuous embedding detector (spot fake reasoning depth)
5. Specificity as quality signal

Deliverables

• Framework cheat sheet (one-pager for your AI team)
• Depth audit worksheet for your current AI deployment
• Access to the formal validation code (Prolog)

Who This Is For

Teams deploying AI agents for:

• Customer service automation
• Code review and generation
• Document analysis and compliance
• Decision support systems
• Any system where AI explains its reasoning

The Research Behind It

Based on original research on in-context particularization - a formal framework for measuring self-referential depth in language models. Validated with Prolog formal reasoning, cross-tested across multiple AI instances, and stress-tested with adversarial analysis.

Pricing

Bundle with AI Readiness Assessment for 15% off.

Book

Contact Maciej or schedule a call.