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 guide Get the templates

~/tests/claude-vs-chatgpt-vs-copilot-broken-budget-spreadsheet-test.md

Claude vs ChatGPT vs Copilot

// hands-on · home

USE

verdict

BrokeCopilot's headline 2024 total was £45 short (656,485 against 656,530) and its dashboard said every sheet had been scanned with no errors

// latest proof

Claude vs ChatGPT vs Copilot: fixing a broken budget spreadsheet

Three AI tools, one broken budget workbook, one sentence of instruction. The test was never whether they could fix it. It was whether they told me what they changed. Real task

Read the test

verdict

Claude wins this test. It was not the smartest tool on every round and its output was not the prettiest. It was the only one that treated what did I change and what am I not sure about as part of the job rather than an afterthought.

broke

Copilot's headline 2024 total was £45 short (656,485 against 656,530) and its dashboard said every sheet had been scanned with no errors

// new

Recent practical notes.

Start with guides

Worksheet26 Aug 2026

The Handover: broken budget spreadsheet AI test kit

A ten-tab budget workbook with 40 deliberate faults, the one-line prompt, and a scorecard. Run it on whatever AI you use and see whether it tells you what it changed.

Test · workUSE26 Aug 2026

Claude vs ChatGPT vs Copilot: fixing a broken budget spreadsheet

I gave Claude, ChatGPT and Microsoft Copilot the same ten-tab programme budget and one loose prompt. All three fixed the error I planted. Only one left a record I would hand to a finance director.

Opinion21 Aug 2026

Claude is adding text watermarks. Here’s what they can actually prove

Anthropic is adding statistical watermarks to text produced by future Claude models. The system can indicate likely Claude involvement, but it can’t establish authorship.

ClaudeAnthropicAI watermarking

GuideWorking

Ask Gemini in Google Chat: what happens to your history

Google is moving Gemini into a new Google Chat surface, but old side-panel conversations will not follow. Here is what changes and what to save.

Ask GeminiGoogle ChatGoogle Workspace

Opinion18 Aug 2026

AI agent production readiness starts with a stop rule

OpenAI paused part of its frontier-model work after a security incident. The lesson for ordinary teams is simple: a successful AI agent demo proves too little unless the pilot has limits, monitoring and a stop rule.

Build · Mixed27 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

// 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.