How to update prompts for GPT-5.5 and modern AI models

Dylan Davis May 2, 2026

• Step-by-step mega-prompts actively limit what modern AI can do
• Binary success criteria help AI self-check before returning output
• Confident hallucinations require proof requirements, not just 'don't lie'

Why AI gives generic results and how a briefing document fixes it

How I Work April 19, 2026

• Generic AI output is a context problem, not a prompt problem
• Let AI interview you by voice to build your briefing document
• Pre-load context into a project so every chat starts informed

How to audit your Claude prompts after a model upgrade

Dylan Davis April 18, 2026

• New models silently break setups that worked perfectly before
• Four checks catch tone, length, vagueness, and tool-skipping regressions
• Smarter models need fewer instructions — but far more precise ones

Automating client data extraction with an AI funnel

Dylan Davis April 11, 2026

• Wrong answers are 3x worse than blank — change AI's incentives.
• Force the AI to quote its source for every extracted value.
• Browser for testing; desktop agents for scale and self-improvement.

Trim your AI instructions to unlock better model performance

Dylan Davis April 7, 2026

• Bloated prompts actively restrain smarter models from performing well.
• 30–50% of rules in most prompts can be deleted without quality loss.
• A five-step quarterly detox removes stale, contradictory, and redundant rules.

How dictation closes the gap between your thinking and AI

Dylan Davis April 4, 2026

• Typing compresses your thoughts — dictation is 4x faster and richer
• The cringe factor fades in three days; most users never go back
• Three tactics: chunk, give AI a job, speak interview answers

Three prompt rules that stop AI from guessing when extracting data

Dylan Davis March 28, 2026

• Smarter AI models guess more confidently rather than admitting uncertainty
• A wrong answer costs 3x more than a blank — change the model's incentive
• Force AI to label every value as extracted or inferred, with evidence

How to force AI to give you genuinely different options

Dylan Davis March 14, 2026

• AI always picks one best answer, then rewrites it multiple times
• Three prompt techniques — MECE, persona rotation, dimension locking — break this
• Subagents eliminate shared context so outputs are truly independent

How to use AI effectively: prompting, tools, and future-proofing

Dan Martell March 5, 2026

• A four-part prompt structure eliminates generic AI output entirely.
• Pull prompting lets AI ask the questions so you do less work.
• Taste, vision, and care are the three things AI cannot replace.

The ChatGPT Skill That Makes You Dangerous in Any Room

Dylan Davis February 21, 2026

A three-rule prompt to stop AI hallucinating on documents

Dylan Davis January 31, 2026

• AI guesses confidently when it can't find something in your document
• Three prompt rules ground the model and cut hallucination rate drastically
• Use self-check, cross-model, or NotebookLM to verify extracted claims

A three-step method to define quality standards for recurring AI tasks

Dylan Davis January 22, 2026

• Your AI standards are trapped in your head — here's how to extract them.
• Binary pass/fail questions eliminate the subjectivity killing your prompts.
• Must-pass questions prevent AI from acing a checklist but missing the point.

Seven official Claude prompting rules most users ignore

Dylan Davis December 23, 2025

• Vague prompts default to average; specificity escapes the model's gravitational pull
• Explaining why you're asking lets the model infer what you didn't say
• Anthropic's research meta-prompt is night-and-day better than standard queries

How to build effective AI prompts and workflows fast

Dan Martell December 9, 2025

• Every prompt needs role, context, command, and format — no exceptions.
• Save perfect outputs as reusable system prompts — that's your future IP.
• Taste, vision, and care are the three things AI can never replace.

AI-generated YouTube thumbnails for $0.12 using a five-step prompt system

Dylan Davis December 4, 2025

• Let AI research best practices and write its own image prompts
• A prompt improver turns a rough brief into a reusable project instruction
• Gemini app vs AI Studio: watermarks vs pay-per-use at $0.12 per image