Practical guides and everything you need to know about AI, automation and digital transformation for small businesses. No unnecessary theory, just results.
Entry 03: May 2026. We added an architect agent for long jobs and hardened handoffs between employees. Every layer of protection brought new ways to fail.
How a small business gets found on Google and in AI answers, built around a real case: our own site lost Google in two days in October 2026.
How a small business turns website visitors into enquiries with web chat and WhatsApp, including the rules, the fees and how we run ours.
A quarter of crawlable pages now carry AI text, yet only one in seven articles ranking in Google is AI-written. What a small business should do with its pages.
Audits of 42 AI chat platforms found 21 leaking conversation titles to ad trackers and 9 leaking your literal prompt. What that means for a small firm.
SMS codes still work everywhere and stop almost nothing. What NIST, CISA and FIDO say about passkeys, how to roll them out in a small firm, and what they leave open.
One study found AI made teams talk to more colleagues. Another found developers working with 79% fewer peers. What decides which happens in your firm.
Firms spend about $2,000 per employee on AI and report 0.2% more productivity growth. What a small business should budget, with the arithmetic for ten people.
Your AI saves hours and takes them back as checking. The measured review burden, two vendor questions that halve it, and a one-week tally for a small firm.
Since 9 December 2026 software sold in the EU is a product and whoever builds it is a manufacturer. What that changes for a small firm, and what it does not.
A randomized trial on 100 live websites put a trained sales agent, a general-purpose model and each site's own chat staff in front of 14.8 million visitors. The agent raised revenue per visitor by 10%. The interesting part is what it does not prove, and which three numbers an owner should measure before believing anyone.
A pre-registered three-month trial with 133 patent lawyers measured what they delivered with an AI assistant, then took the assistant away. Output went up for everyone. Durable judgement went up only for the people who already had it, and the juniors split into two groups.
