By the PrototypeVerdict Test Team ยท Last updated: October 29, 2026
We may earn a commission if you buy through links on this page. How we make money.
This page explains how we decide what goes on PrototypeVerdict, where our facts come from, how we keep pages current and how we fix mistakes.
Independence
- Rankings aren’t for sale. No company can pay for a review, a higher score, a better verdict or a place on one of our lists. We don’t accept money, free upgrades or other perks in exchange for coverage.
- Affiliate status doesn’t change scores. A tool that may pay us a commission gets the same three test builds, the same fix-message limit, the same checks and the same scoring rubric as a tool that pays us nothing. Bolt.new, for example, pays us nothing and was tested the same way as Lovable and Base44.
- We pay for what we test. We buy the paid plans we test on. When we test on a free plan, we say so and don’t present the results as covering paid plans.
- We publish the bad parts. Every review lists what broke, what we haven’t tested yet and who should skip the tool, whether or not the tool can pay us.
How we make money is explained in our affiliate disclosure.
Where our facts come from
We use two kinds of sources:
- Each tool’s own official pages: its pricing page, documentation, help center and published terms. We link the page next to the fact and show the date we checked it.
- Our own test logs. Times, credit and token counts, what worked, what broke and our screenshots all come from the test sessions recorded in our logs. The method is on our how we test page.
When we can’t confirm something on an official page or in our own tests, we either leave it out or say clearly that it’s unconfirmed. If we ever rely on another source, such as a third-party report, we name it and say that it isn’t from the tool itself.
How we keep pages current
- Prices are rechecked every month for every tool we’ve reviewed, against the tool’s own pricing page. Every review shows a “Pricing checked” date.
- A full retest every quarter. We rerun the three standard builds in every tool we cover once a quarter, and sooner when a tool ships a major release that could change our results. Every review shows a “Last tested” date.
- When something changes, such as a price, a plan, a feature or an affiliate program’s terms, we update the affected pages and the affiliate disclosure.
- Dates mean real changes. We change a page’s “Last updated” date only when we’ve actually changed something on it.
Corrections
If you think we got something wrong, email contact@prototypeverdict.com. Tell us the page and what you think is wrong, with a source if you have one.
When we confirm a mistake, we fix it on the page and add a dated note to that page’s update history saying what was wrong and what we changed. We don’t quietly rewrite a test result. If a retest gives a different result, we show the new result with its date.
Update history on every page
Every page ends with an “Update history” section. It lists each meaningful change with its date: first publication, new prices, new test results, new screenshots and corrections.
How we use AI agents and AI tools
Our test sessions are run by AI agents working in real accounts on the plans we bought (Bolt on its free plan). The agents follow our fixed build prompts, and every result is logged and reviewed before it goes on a page. Every time, credit count, result and screenshot we publish comes from those logged sessions.
AI tools also help us draft, edit and check pages, including checking facts against our logs and sources. Prices come from each tool’s own pricing page on the date shown. Nothing is published without review and approval.
Questions
Questions about this policy go to contact@prototypeverdict.com. You can also read about PrototypeVerdict or go back to the homepage.
Update history
- October 29, 2026: First published.