>_clıffinkent
Cliff, who helps working people get more useful work from AI
// tested in public

Practical AI for work, home and play.

Cliffinkent helps working people get more useful work from AI. Methods you can copy, tested against real jobs with the broken bits left in.

Start with a guideGet the templates
~/tests/claude-fable-5-vs-gpt-5-6-project-handover.md
Claude Fable 5 and GPT-5.6
// hands-on · work
USE
verdict
WorkedBoth models found the ten planted facts and two contradictions.
NoteClaude Fable 5 connected governance conflicts across separate sources.
BrokeClaude Fable 5 needed more editing before its output could be used as a steering pack.
// latest proof

Claude Fable 5 vs GPT-5.6: Project Handover Review

I gave both models the same messy 20-file project handover. They tied on the main score, but GPT-5.6 was the better first choice for regaining control while Claude Fable 5 was the stronger reviewer.

Read the test
verdict

GPT-5.6 was my better first choice for taking control of a messy project. It produced a clearer, more steering-ready handover pack with less rework. Claude Fable 5 was the stronger second reviewer when disputed approvals, weak governance records or conflicting evidence needed closer scrutiny.

worked
  • Both models found the ten planted facts and two contradictions.
broke
  • Claude Fable 5 needed more editing before its output could be used as a steering pack.
// new

Recent practical notes.

Start with guides
Build · Home27 Jul 2026

How I shipped a real app with Claude Code and Codex

I used Claude Code and Codex to turn a weekend prototype into a live iPhone app. The hard parts were evidence, billing, testing and store review.

Claude CodeCodexAI coding
Test · workUSE22 Jul 2026

Claude Fable 5 vs GPT-5.6: Project Handover Review

GPT-5.6 produced the better steering-ready handover. Claude Fable 5 was stronger at challenging disputed approvals and conflicting evidence.

Prompt09 Jul 2026

Steering Committee Test Pack: run the AI deck test yourself

The exact prompt, ten files of realistic project mess, a locked committee deck and the answer key to ten seeded traps. Run my AI deck test on your own tool.

Test · workWATCH09 Jul 2026

ChatGPT vs Copilot in PowerPoint: the steering committee test

I gave ChatGPT and Copilot ten files of project mess and a locked committee deck. One edited the template perfectly. One signed my name to it. See the scores.

Test · workUSE01 Jul 2026

Claude Sonnet 5 test: one call, two clean outputs

I gave Claude Sonnet 5 one messy transcript and asked for two clean outputs in one prompt. Read what worked, what broke, and whether it holds up on day one.

Opinion19 Jun 2026

Most AI demos skip the bit that matters

A demo is built to succeed. The part that decides whether a tool is any use, how it fails on a real task and whether you would notice, never makes the cut.

// the weekly note

One useful AI note a week.

A guide, template, build, test or opinion, with the useful bit and the broken bit left in. Sponsorship and free access are disclosed where they apply.

// practical AI for work, home and play. unsubscribe whenever.