Practitioner Guides
Operational guides for professionals who work with AI systems. Each guide addresses a specific structural problem, explains why it happens, and provides a practical procedure you can use immediately. Drawn from the Synthience research corpus and written for practitioners, not academics.
New articles in full, plus a short note on each new paper, by email: Subscribe on Substack →
published
Start Here
If you only read one of these guides, read this one. Ten short, concrete practices that prevent the most common ways AI-assisted work goes wrong. Each one is one Problem, one Solution, and one example prompt you can copy directly into your next AI session. No technical background required, works on any AI platform, applies to any kind of task.
Format filter and search work together. Results match all active selections. Use “All Articles” view to see everything in a single list.
View:
Sort:
- All Practitioner Guides
- PG-000: 10 Things Every AI User Should Do
- PG-002: AI-Assisted Editing Without Silent Loss
- PG-003: Verify Before You Work: A Minimal Sequence for AI Document Ingestion
- PG-006: The Ingestion Verification Protocol: Does Your AI Actually Read Your Documents?
- PG-004: You Are Accepting the First Adequate Answer
- PG-005: Your AI Updated the File. Did It Preserve What It Didn’t Touch?
- PG-001: How to Work Reliably With Conversational AI Over Time
- PG-007: Introducing the Synthience Institute: The Missing Middle in AI Safety
- PG-008: The Continuity Architecture Module
- PG-009: Make the AI Show You the Source
- PG-010: Don’t Trust What the AI Says About Its Own Work
- PG-011: The Cross-AI Adversarial Review Protocol
- PG-012: Make the AI Tell You What It’s Guessing
- PG-013: You Don’t Have a Model Problem. You Have a Method Problem.
- PG-014: You Don’t Have a Method Problem. You Have a Routing Problem.
- PG-015: You’re Not Paying for the Model. You’re Paying for the Conversation.
- PG-016: Checking the Checkers: How Four Verification Protocols Were Corrected in Public