How we write and recommend

AI do it aims to make software and AI easier to understand through clear explanations, practical tasks and honest limitations.

Evidence before claims

Jay has tested the tools covered in our reviews and guides. We separate his observations from provider documentation, and use fictional records for worked examples. Specific findings, scores and test details are added as each review is edited. We link to primary sources where practical. Check current features, prices and terms before buying.

A consistent review rubric

Scored reviews use price, ease of use, UK tax fit, bank feeds, invoicing, integrations and support. Scores run from 0 to 5 in half-point increments and require evidence and a reason. An irrelevant dimension is marked not applicable with an explanation. UK tax fit is not a certification.

We only publish an overall score when at least four applicable dimensions have evidence. It is their unweighted mean, rounded to one decimal. Research notes without evidence remain unscored.

Commercial relationships

Affiliate links are labelled before you use them. We may earn a commission if you buy through links on a page. It doesn't change what we recommend. Paid placements begin with “Sponsored” and remain distinct from independent editorial content.

Read the affiliate disclosure.

Corrections

If something is unclear or out of date, email Jay with the page address and the detail you want us to check.