How we review
ReviewRamble is a directory and a review site: everything we list gets a sourced, rated review. A review is only as good as the evidence behind it, so this page sets out where ours comes from, how we score, what we list, and the limits of what we claim.
Independence
ReviewRamble is funded by Google advertising. We take no commission on links and no payment for coverage, and advertisers have no say in what we review or how we rate it.
Sources
Each review draws on at least two sources: the product’s own site plus its pricing page, documentation, changelog, licence or reputable independent coverage. Every price, limit, date and benchmark in the text comes from one of the sources listed at the foot of the page. Where a vendor contradicts itself, we say so. Where something is not public, we say that too rather than guess.
Benchmark figures published by a vendor about its own product are labelled as the vendor’s claims. We prefer independent leaderboards where they exist.
How entries are written
Entries are researched and drafted with AI tools, working only from the sources listed at the foot of each page, and published under the ReviewRamble editors byline. Automated checks reject an entry that lacks its sources or dates, uses promotional language, or whose translations drop a figure from the English.
Dates
Every review shows the date its facts were last checked. AI products change quickly; treat anything older than a few months as a starting point and confirm prices on the vendor’s site before you buy.
Hands-on testing
Unless a review says otherwise, it is a desk review: written from documentation, source code, pricing and user reports rather than from our own tests. We are adding hands-on notes, timings and screenshots, starting with open-source projects and local models, and those reviews will say so explicitly.
Ratings
- 4.7 and above: the clear leader in its category. Rare by design.
- 4.0 to 4.6: strong. Recommended for the people the review names.
- 3.5 to 3.9: fine, with real trade-offs you should understand first.
- Below 3.5: hard to recommend over the alternatives.
A rating compares a product with its current rivals, so scores can fall when something better arrives, even if the product itself has not changed.
What we list
The directory covers AI tools, models, MCP servers and open-source projects. To be listed, something has to be available now, do something distinctive rather than wrap another product, and, for open-source projects and MCP servers, have had a release or commit in the past twelve months. We do not list products that have shut down; when a listed product closes or a project goes quiet, we say so on its page or remove it.
Directory listings that we have not yet reviewed appear in search and on directory pages, but do not get a page of their own. To suggest something for the directory or report an outdated entry, use our submission form.
Corrections
When we get something wrong, we correct it and update the review’s checked date.