Blog
Working notes on AI implementation, workflow automation and internal assistants, written from operational experience with every source listed.
Most writing about AI implementation is either vendor marketing or a benchmark table that is out of date within a month. Articles here deal with the decisions a business has to make. Whether a process is worth automating, where data goes when you call a model API, what a document set needs before retrieval works, and what happens when an integration fails unattended.
Every factual claim is checked against a primary source, and the sources are listed at the end of the article with the date they were read. Where something could not be verified, it says so rather than filling the gap with a plausible number. Articles about fast-moving subjects carry a visible Last verified date so you can judge how much to trust them.
There are no client case studies here, and no invented metrics. What sits behind the writing is the operational experience of Apefo Ltd's own businesses: hosting operations, website delivery, SEO and content workflows, and the CRM and reporting handoffs between them.
No UK law requires a business to have an AI policy. Data protection law does require you to handle personal data properly whether AI is involved or not. Here is a policy short enough that a twenty-person company will adopt it, with the obligations marked.
Read the articleArticle 50 transparency obligations apply from 2 August 2026. When a UK business is caught, what it must disclose, and where the line between deployer and provider sits.
Zapier, Make and n8n list prices verified on 22 July 2026, what each vendor counts as a billable unit, and the arithmetic for finding your own crossover point.
The famous AI failure statistics do not survive checking. An eight-factor test for which processes to automate, which to fix first, and which to leave alone.
Kimi K3 is a 2.8-trillion-parameter model and Moonshot recommends 64 or more accelerators to deploy it. What that means if you were hoping to run it on your own server.
Next step
Describe one task your team repeats and you get a straight view on whether it is worth changing.
