Andrew Mayne Prompts

Search posts

Navigation

Home
Posts
Chronology
Reference index
About

Part I — Origin Story and Base-Model Prompting

How I Became OpenAI's First Prompt Engineer
The Evolution of Prompts: From Completion to Systems
Base Models vs Post-Training: What Each Layer Does
Model Identity and Statelessness: Why Explicit Context Matters
The Stateless AI Guessing Game: A Prompting Lesson in Memory
GPT Demo Set List: Early Prompt Patterns That Still Hold Up
Early Sentence-to-Email Prompts: A Foundational Transformation Pattern
GPT-3 Emoji Story Demo: Narrative Compression in Tokens
Separating Instruction from Content: A Core Prompt Reliability Pattern
Magic Words in Prompting: Domain Terms That Steer Behavior
Invoking Experts in Prompts: When Persona Framing Improves Results
GPT-3 Grammar and Style Editing in Practice
Magic Phrases for Moderation: Prompt Patterns That Improve Safety Calls
GPT-3 for Regex, Bucket Policies, and Solidity Tasks
Rethinking best_of in GPT-3: Why It Misleads
The Fifth-Grade Summary Moment: Audience-Aware Compression
Interlude: Prompting as pattern design (why formatting beats adjectives)

Part II — Reliability and Control Patterns

Mini Prompts for Trick Questions and Nonsense Inputs
Prompts to Reduce Hallucinations: Practical Control Patterns
Cross-Temperature Hallucination Testing for Sanity Checks
Temperature in LLMs Explained: What It Actually Controls
Seeded Creativity for LLMs: Controlled Randomness That Helps
Creating Better Quiz Distractors with LLMs
Bracketing Letters for Wordle: Token-Level Prompt Control
The Missing Bracket: How Tiny Formatting Errors Break Outputs
Prompt Repetition and Rephrasing: A Reliability Tactic That Lasts
Interlude: When prompt engineering became an engineering problem

Part III — Efficiency: Cost, Compression, and Small Models

Prompt Size Reduction Checklist: Cut Tokens Without Losing Quality
Small Model Advantages: When Smaller LLMs Outperform Bigger Ones
Small Models, Big Knowledge: Prompting Past the First Guess
Using Small Models for Complex Natural-Language Tasks
Large Text Pattern Analysis with Prompted Models
Prompt Maker: How to Teach Prompt Patterns by Example
Compute at Scale: Growth, Limits, and AI Demand
How Small Can AI Be? Practical Limits and Opportunities
Interlude: The cost curve changes everything

Part IV — Context and Scaffolding Workflows

Context as an AI Lever: The Compounding Effect of Longer Windows
Small Capabilities, Big Ramifications in Prompt Design
Scaffolding Long-Form Content: Prompt Patterns for Coherence
Radio Play Scaffolds: A Better Prompt Pattern for Story Generation
Building AI Choose-Your-Own Adventures with Prompt Scaffolding
Character-Threaded Summarization for Long Documents
Memory in Conversational AI: Why Context Persistence Matters
Outcome-Oriented Prompting: Define Success, Then Generate
Style Guides for AI Writing: Getting a Specific Voice
Crystallized vs Fluid Intelligence in Language Models
Personal AI Evaluation Methods for Real-World Quality
Interlude: From ‘write me a thing’ to outcome specs + evals

Part V — Retrieval and Grounding Systems

Understanding Embeddings for Better Prompting and Retrieval
Embedding-Based Retrieval Strategies That Actually Work
Context vs Retrieval: A Practical Decision Framework
Grounding Prompts with Wikidata and SPARQL
The Prompt Context Flywheel for Continuous Improvement
Interlude: Prompting becomes a system: retrieval + transformation + reuse

Part VI — Fine-Tuning as an Adaptation Layer

Fine-Tuning Fundamentals: When to Use It and When Not To
Fine-Tuning Methods Guide: SFT, DPO, and Beyond
Cost Savings via Fine-Tuning Smaller Models
Model-Assisted Data Preprocessing for Better Fine-Tuning
Interlude: The prompt is now a workflow spec

Part VII — Products, Multimodal Systems, and the Frontier

GPT Tools: Fast Prototypes, Real Constraints, and Shipping
Discovering Useful Libraries with AI Coding Prompts
Code Refactoring with GPT-3: Practical Prompt Patterns That Work
Tool Makers vs Tool Users: Where Product Value Actually Lives
Lessons from an Ambitious AI Build
Why I Didn't Launch AI Channels
Hackathons and Model Capabilities: What Fast Experiments Reveal
GPT-4 Vision Refrigerator Demo: A Practical Multimodal Moment
Localization Techniques for Vision Models in Real Workflows
Vision Models at the Frontier: What Changed and Why
Big and Small Models in Robotics: A Hybrid Architecture
The Uneven AI Frontier: Why Capabilities Arrive Jagged
The Frontier Is Wider Than It Looks
Challenging AI Paper Claims with Practical Replication

Reference index

History and base-model era
Scaffolding and long-form coherence
Reliability / hallucinations / self-checks
Formatting and “interface fixes” (schemas, separators, brackets)
Cost / small models / throughput
Retrieval / embeddings / grounding
Tools / agents / product workflows
Vision / multimodal / robotics
Evaluation and mental models

Evaluation and mental models

Outcome-Oriented Prompting: Define Success, Then Generate
Crystallized vs Fluid Intelligence in Language Models
Personal AI Evaluation Methods for Real-World Quality
Fine-Tuning Fundamentals: When to Use It and When Not To