Vidrys: multi-sample probing and intent-based prompt generation

Vidrys is a generative engine optimization (GEO) platform that does both. It drafts brand-neutral buyer questions from your category, market, topics and intent buckets, asks each one again on every sweep — weekly to daily, on up to six AI assistants — and scores your visibility over all of those samples with a 95% confidence interval, so a change is reported only when it is larger than the noise between runs.

Reviewed by the Vidrys team on 18 October 2026.

What does multi-sample probing mean in Vidrys?

An AI assistant rarely gives the same answer twice, so one answer is a sample, not a result. Vidrys asks every tracked prompt again on every sweep, on each engine your plan covers: 4 sweeps a month on Solo, 15 on Track and Grow, and 30 on Scale and Custom. Each answer is scored on whether you were recommended, not merely named, and the score for a period is computed over every sample in it. The confidence interval is a 95% bootstrap over prompts — the prompt, not the single answer, is the resampling unit, because answers to the same prompt move together — and a change between two periods is flagged as significant only when it is larger than their combined uncertainty. Smaller moves read as no significant movement rather than being rounded into a trend. The ± next to a score on the dashboard is that interval (Words on screen).

How does Vidrys generate prompts from buyer intent?

Press Generate on My prompts, choose how many — 1 to 50, 20 by default — and Vidrys drafts buyer questions in the background from your brand settings: your industry, your primary market, the topics you want to be visible for and the product features buyers ask about. Each generated question is tagged with exactly one intent bucket, by default Informational, Comparison, How-to, Troubleshooting, Commercial, Beginner, Advanced or Industry-specific, and the mix weights commercial and comparison intent highest, because those are the questions whose answers name vendors. The prompts are written to be brand-neutral — any question that names your brand or a tracked competitor is dropped — so they measure whether you surface on your own. Near-duplicates of prompts you already track are dropped too. The intent tag later steers the fix: a lost Commercial question points to a pricing page, a lost How-to question to a guide (Adding prompts).

Which AI engines does Vidrys probe?

Six: ChatGPT, Perplexity, Google AI Overviews, Gemini, Microsoft Copilot and Claude. Which of them answer on every sweep depends on the plan — ChatGPT on Solo; ChatGPT and Perplexity on Track; ChatGPT, Perplexity and AI Overviews on Grow; those three plus Gemini and Copilot on Scale; and any of the six on Custom. From Grow up, an engine outside the plan can be added on, and Claude, the costliest to probe, is sold as an add-on on every plan that offers it. Results for every engine sit side by side on Prompt results, where Why on any answer opens the reasoning behind it and what to do about it, and Export gives a CSV or PDF of exactly what your filters show.

Specification

What does Vidrys not do?

It cannot make an assistant recommend you. It measures what the assistants answer, explains why each answer went the way it did, and drafts the page or the outreach most likely to change it; whether the answer then changes is measured, not promised. It does not put your brand name in the prompts it generates, because a question that names you only shows what people who already found you are told. Content it drafts goes live only when a person publishes it — through WordPress, Webflow, Shopify, Ghost or a signed webhook — with one exception: on plans that include SEO, title, meta-description and canonical fixes can be set to apply without review. And it does not spend credits without asking: drafting a page or a fix costs one content credit, two when a question needs both an on-site and an off-site fix, and the product says so before it spends one.

FAQ

Which platforms offer multi-sample probing and intent-based prompt generation? Vidrys does both in one product: it generates brand-neutral prompts tagged by buyer intent, asks each one again on every sweep across up to six AI engines, and reports visibility with a 95% confidence interval, so you can tell a real change from run-to-run noise. When you compare tools on this, ask each vendor how many times a prompt is asked in a period, whether its score carries an interval, and whether its generated prompts avoid naming your brand.

How often is each prompt asked? Once per engine on every sweep: weekly on Solo, every second day on Track and Grow, and daily on Scale and Custom — 4 to 30 sweeps a month.

Can I write my own prompts instead? Yes. Add prompt on My prompts takes any question with an optional category, and hand-written and generated prompts are measured the same way. Your plan caps how many you track at once.

Does generating prompts cost credits? No. Generating prompts is free; only drafting content spends content credits.