The tools behind every review · built and run by RankDude
Checkers: the tools I test with
A grade is only as good as the tool that produced it. These are the timers, probes, recorders and answer keys used during every hands-on review, each explained on its own page: what it does, how it works, what it records, and what it cannot catch.
- Checkers
- 27
- Categories
- 5
- Reviews they built
- 52
AI agent
- After every taskOutput checkerOpens the file or answer an agent produced and scores every checklist point against the answer key written before the test.Built, not used in a published review yetHow it works
- After the output checker, once per reviewFalse-done counterCounts the tasks an agent called done while a checklist point was wrong and the agent did not say so.Built, not used in a published review yetHow it works
- Before and after every taskCredits readerReads the product's own credits meter before and after each task and writes the numbers to the task ledger.Built, not used in a published review yetHow it works
- Before a review is published, and again after every editLedger verifierRecomputes every number in a review from the task ledger and the credits meter and lists what does not match.Built, not used in a published review yetHow it works
- During a task, when the video needs itRun recorderRecords the screen while an agent works on a task, with a clock, for the review video.Built, not used in a published review yetHow it works
- After the test, once per reviewInjection checkerChecks whether a hidden instruction planted in a page, file or calendar was followed, ignored silently, or ignored and reported.Built, not used in a published review yetHow it works
- Live throughout a testSafety canaryA visit counter that records when an agent opens an address planted in a hidden instruction, so a silent follow is caught too.Built, not used in a published review yetHow it works
AI app builder
- After the first build and after every changeSpec checkRuns the written checks of the Fitting Room brief against the live app every builder made, and counts what passes.Used in 20 reviewsHow it works
- After the first buildAccess probeReplays the app's own data requests with the wrong user or role and records whether the builder's app blocked them.Used in 19 reviewsHow it works
- Around every prompt of the buildCredit ledgerLogs every prompt sent to the builder with the credits, tokens or money its own meter showed before and after, and what came back.Used in 21 reviewsHow it works
- After publishingPage speed testRuns Google Lighthouse's mobile test on the published app or site three times a page and grades the median score.Used in 21 reviewsHow it works
- Whenever a free plan blocks a stepPaywall logWrites down every wall a free plan or trial hits: the exact words on screen, a screenshot, and the price to pass it.Built, not used in a published review yetHow it works
AI receptionist
How ai receptionist are tested
- Every call of the testTest callerThe same caller phones every receptionist with the same calls, answers its questions from one fixed persona, and records the whole call.Used in 10 reviewsHow it works
- Every turn of every callReply timerMeasures how long the receptionist takes to pick up and, on every turn, the gap between the caller finishing and the receptionist's first sound.Used in 10 reviewsHow it works
- During and after every callFact and trap checkChecks every fact the receptionist gave against the business brief, and whether it held its ground on the questions written to trip it.Used in 10 reviewsHow it works
- After the message-taking callsMessage checkAfter a call that leaves a message, reads what the receptionist delivered and when, and checks every field against what the caller said.Used in 10 reviewsHow it works
- The two transfer calls of the testTransfer checkAsks for a named person during and after business hours and checks whether the receptionist transfers to the right person, what it asks first, and what happens when nobody answers.Used in 10 reviewsHow it works
- The three booking calls of the testBooking checkBooks, moves and cancels an appointment by phone against a prepared calendar, and checks the calendar afterwards to see what really changed.Used in 5 reviewsHow it works
- During setup, for vendors that read a websiteTest business pageA hidden web page for the fictional plumbing business, so receptionists that learn a business from its website all read the same facts.Built, not used in a published review yetHow it works
- After the calls, before the review is writtenCall clips and transcriptsTurns each recorded call into the listening copy and the written transcript the review shows, without touching the timing record.Built, not used in a published review yetHow it works
AI website builder
How ai website builder are tested
- After publishingPage speed testRuns Google Lighthouse's mobile test on the published app or site three times a page and grades the median score.Used in 21 reviewsHow it works
- After the first draft of each siteFact checkChecks the sixty facts of the three briefs on the live sites a builder made, page by page, and lists what the builder invented.Used in 3 reviewsHow it works
- After the first draft of each siteDesign and SEO checkerOpens every page at phone, tablet and desktop width and lists layout faults, contrast failures and what the published code gives search engines.Used in 3 reviewsHow it works
- After the first draft of each siteFixed task listsTwelve edits, twelve business features and three support questions, the same for every builder, each timed and marked done or not.Used in 3 reviewsHow it works
- Whenever a free plan blocks a stepPaywall logWrites down every wall a free plan or trial hits: the exact words on screen, a screenshot, and the price to pass it.Built, not used in a published review yetHow it works
AI girlfriend
- During every chat conversation of the testChat timerTimes every chat reply from pressing Enter to the first words appearing, over the same conversations on every app.Used in 18 reviewsHow it works
- Inside each chat conversation of the testMemory testGives the companion a made-up name and three facts, then asks for them back twice, later in the conversation.Used in 18 reviewsHow it works
- During every voice call of the testCall recorderRecords voice calls on both sides and measures, on the recording, how long the companion takes to start answering.Used in 13 reviewsHow it works
- After every recorded callVoice checkRates how natural a companion's voice sounds: sound quality, rise and fall, pace and pauses on every recorded answer, then listened to.Used in 16 reviewsHow it works
Not checkers, but close
Live monitorsUptime, page speed, security and prices, measured every day after a review is published, plus the monitors that keep an AI agent review fresh.Data sourcesThe outside data behind the monthly statistics pages, each explained: what it measures and what it cannot tell me.
Questions
What is a checker?
A tool RankDude uses during a review to test, measure or mark a product: a timer on every reply, a probe that replays an app's own requests with the wrong user, an answer key a file is scored against. Each runs the same way on every product in its category, and the grade comes from what it records.
How is a checker different from a monitor?
A checker runs during the test and stops. A monitor keeps measuring after the review is published: uptime, page speed, security and prices every day. The monitors have their own section.
Are the figures on these pages typed in?
No. Every figure shown under a checker is read from the review that used it, at build time, so it is the same figure the review shows.
Can a brand run its own checker or change a result?
No. The tools, the test accounts and the data are mine, every brand gets the same tools and the same lists, and a result is never edited by hand without the review saying so.
Written by Miles Foster, Owner and tester, RankDude.