Predictika
Resources
Papers, patents and talks
Organized by who is reading. Start with the first list if you are deciding whether this matters to you.
For decision makers
- Safe AI May Be a Myth. Checked AI Is Not.
- Don’t Trust AI. Verify Its Actions.
- Personal Agents Are Going Shopping. Who Checks the Cart?
- You May Not Realize It, But Your Business Has a Configuration Problem
- The Business Case for Constraint-Based Configuration
- Accuracy Is Not Consistency: Why Enterprise AI Needs Deterministic Guardrails
- Why Large Language Models Fail at BNPL Logic, and How a Deterministic Co-Processor Can Fix It
- Pricing Is a Configuration Problem
- Beyond SteelBrick: A Constraint-Based Path Forward for Salesforce CPQ Customers
For agent builders
- FAQ: How to stop AI agents from taking the wrong action
- Syntactically Valid, Semantically Wrong
- ExecGuard: Execution Integrity for Enterprise AI Agents
- Your Agent Shouldn't Hold Your Wallet, or Anything Else of Value for That Matter
- Guiding AI Agents to Reason Correctly
- Conversational LLMs Guided by a Logic Engine for Accurate and Reliable Reasoning
- Where is the AGI in LLMs if they cannot cross the river?
Patents
US 11,544,475 B2
Core patent on deterministic validation of LLM-driven systems.
US 11,914,970 B2
Extended coverage for the logic validation engine and its hybrid architecture.
Read the patent releasePatent pending
Conversational autonomous AI agents guided by a logic engine for accurate and reliable reasoning.
Listen
Why AI Agents Fail at CPQ, and How to Fix It
Dr. Sanjay Mittal on the CPQ Podcast with Frank Sohn: why "95% correct" is a failing grade in enterprise AI, and why prompt engineering cannot fix it.
Listen to the episodeLLM structural mismatch and enterprise risk management
Industry perspectives on the DeepSeek revolution.
Listen to the audioTry it and press coverage
EU flight-delay compensation bot
An LLM front end with the deterministic engine deciding what the regulation entitles you to.