Notes from
the work.

What I learn building automation and software for businesses. 233 posts, free to read, no sign-up.

The AI Implementation Opportunity Report 2026

12,000 words and 80+ verified sources: pricing benchmarks, opportunities by sector, GEO trends and the root causes of small-business AI failures that most reports miss.

Read the report
Sort Newest Oldest Longest read
All 233 AI agents and tools 31 AI strategy 75 Automation 68 Software and development 37 UK business and governance 22

Showing 1 to 4 of 4 posts

AI agents and tools6 Apr 20268 min read
Meta’s AI Safety Director Lost Control of Her Email Agent. Here’s What It Means for Your Business Summer Yue, Meta's head of AI alignment, watched her OpenClaw agent delete 200+ emails while ignoring her stop commands. The agent didn't go rogue, it forgot her instructions. If it can happen to the person in charge of AI safety, it can happen to anyone. Here's what you need to prevent it.
AI strategy18 Feb 202617 min read
Claude Sonnet 4.6 vs. Claude Opus 4.6: What the Safety Evaluations Tell Us That the Benchmarks Don’t Benchmarks don't tell the whole story. The safety data reveals why Sonnet 4.6 may be the better choice for autonomous agents in real-world environments, despite Opus 4.6's higher raw capability.
Software and development18 Feb 20268 min read
GPT-5.3-Codex’s Cybersecurity Risk Rating: What OpenAI’s Own Safety Team Is Worried About OpenAI classified GPT-5.3-Codex as high-risk for cybersecurity capabilities: a first for any model. What the red team found, how OpenAI is controlling it, and what it means for your infrastructure.
AI strategy18 Feb 20266 min read
What Claude Opus 4.6’s System Card Actually Says About AI Safety Anthropic's detailed safety analysis of Claude Opus 4.6 reveals uncomfortable truths about model behaviour during deployment. Here's what your team needs to know before adopting it.

Want this done for your business?

One paragraph is enough. I reply personally.

Start a project