When your customer asks AI for a company like yours, does it name you, or a competitor?

CitePulse is an AI citation audit and monitoring platform operated by GeoFlow Solutions Marcin Gamża (Your Application Intelligence Partner). It measures whether AI assistants, ChatGPT, Perplexity, Gemini, Claude, Grok and Google AI Overviews, cite your company when customers ask real buying questions, flags when AI states something false or out-of-date about you, and names the competitors recommended instead.

The free mini-audit covers ChatGPT, Perplexity and Gemini, with Google AI Overviews read as a source. The one-time Full Audit adds Claude, and the Protect and Scale monitoring plans add Grok (xAI), up to five assistants in total. Google AI Overviews is read at every step as a source, not counted as one of the engines.

SEO tools analyse your page source and Google rankings. CitePulse queries the AI assistants directly with the questions your customers actually ask, and reads the real answers: whether your domain is cited, what AI says about you, and whether it is accurate. Rankings measure a page users increasingly skip; citation share measures the answer they read instead.

Citation share is the percentage of AI answers to a fixed set of buyer questions in your category where your domain is cited as a source, relative to all cited domains, per engine, tracked over time. It is the share-of-voice metric for AI search, it measures the answer users actually touch, not the impressions and clicks on a page beneath it. We explain it in full in Citation share: the new share of voice.

Fact Integrity lets you approve the facts AI must get right about your brand, prices, key numbers, claims, and monitors whether AI answers, your own website and the third-party listings AI cites still state them correctly. It flags “drift” when AI contradicts an approved fact and “stale echo” when AI repeats a value you already retired, pointing to the source to fix. It is included in every monitoring plan (Shield 10, Protect 20, Scale 30 approved facts) and measures consistency with the facts you approved, it never controls what AI says.

The mini-audit is free, with no signup or card. The one-time Full Audit is $39 plus $9 per extra service area, and monitoring plans are Shield $89, Protect $169 and Scale $349 per month. Every monitoring plan includes Fact Integrity.

About 30 seconds. Enter your website and pick your service areas, CitePulse runs technical checks plus real buyer questions against ChatGPT, Perplexity and Gemini (with Google AI Overviews as a source), then shows an AI Reputation Score from 0 to 100 with the competitors cited instead of you. The full PDF report, with copy-paste Auto-Fix files, is emailed to you.

The AI Reputation Score is a 0–100 measure built from four stacked layers: Readable (can AI crawl your site), Understood (does AI grasp who you are), Cited (does AI actually name you when buyers ask, the money layer), and Quotable (is there source-backed material AI can lift). A weak lower layer caps the ones above it, so the score points you at what to fix first.

Yes. The Accuracy Audit compares what AI assistants say about you against the real content of your site and flags hallucinations, outdated pricing, mis-described services or false claims, with exact quotes. Fact Integrity extends this to the facts you formally approve, checked across AI answers, your own site and third-party listings on every monitoring cycle.

No, and neither can anyone else. No AI engine publishes how it weights sources, and citations shift from day to day with retrieval and model temperature. CitePulse sells measurement and fixes, not guaranteed citations: we improve the signals AI reads, technical readiness, clear claims, source presence, and track your citation share so you can see whether a change actually worked.

Yes. The Scale plan ($349/month) covers up to 10 client domains from one dashboard, with white-label branding on client-facing reports and API access. Agencies use it to add AI citation audits and monitoring to existing retainers under their own logo.

For each buying question, CitePulse classifies the sources the AI engines actually cited. Door A: ordinary pages win (or you are already cited), so your own page can compete on that question. Door B: at least two thirds of the cited sources are best-of lists, directories or marketplaces, so presence on those third-party pages is the lever, not rewriting your own page. The rule is published as DOORS-1.0 on the methodology page. CitePulse never claims you are absent from a page; it shows which pages win.

You mark a change as done in the Action Center (or CitePulse detects the page change itself). The change gets a date and a baseline of up to three earlier runs of the same frozen questions. Later monitoring runs are compared per question and engine, and each cell gets one of three states: rose beyond the engine’s noise band, fell beyond it, or within normal variation. A result counts as observed only after at least two comparable runs, six repetitions per cell and 14 days. This is correlation with a date and a band, never proof that the change caused it.

A lot: published 2026 measurements put the day-to-day turnover of cited sources at roughly two thirds, and same-day repetitions agree only partially. CitePulse measures this on its own daily panel (NOISE-1.0): the same five buying questions, four engines, three repetitions, every day. The noise band, repetition agreement and day-to-day source overlap per engine are published on the methodology page and as JSON at api.citepulse.ai/api/public/noise. Until an engine has 14 days of data, a default band of 0.34 is used and stated.

A single run is a snapshot, not a state. Instead of “AI cites this page”, the dashboard reports in how many of the last comparable runs (the same frozen questions and the same measurement version) each of your pages appeared among the cited sources, for example 2 of 3, with the engines that cited it.

Because Microsoft offers no official API for the consumer Copilot. The documented Copilot APIs cover Microsoft 365 Copilot inside a licensed tenant, and Azure’s Grounding with Bing Search generates our own model’s answer rather than showing what Copilot says. Tools that claim to track Copilot automate a browser session, which produces results that cannot be repeated or audited. CitePulse shows Copilot in the question × engine matrix as “not measured” and will add it the day official programmatic access exists. Checked 2026-09-06.

Yes. Every audit generates five buying questions grounded in your own site and freezes them as a comparable series. On monitoring you add your own on top: up to 10 on Shield, 20 on Protect and 30 per client domain on Scale. Every weekly run asks the series plus your questions, three times per engine, and a question added later is reported only against the runs that asked it.

Which pages the AI crawlers actually read: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot, Googlebot, Bingbot, Applebot and others. You connect your Cloudflare zone with a read-only analytics token (stored encrypted, removable in one click) or upload a log file; only day, crawler and path aggregates are kept, never IP addresses. The dashboard crosses that with the pages AI cites (read and cited, read but not cited, cited but rarely read) and with robots.txt. Correlation, never causation.

No articles and no marketing copy. Draft Studio assembles three structural blocks for a page that should win a question: an FAQ block with FAQPage schema, a comparison-table skeleton with your column filled and competitor cells left empty, and a visible last-updated line. It runs without a language model: every sentence comes from your approved fact registry and a missing fact is marked as a gap, never guessed. You paste it, mark it done, and the Action Center measures what happened next.

Yes, in the dashboard under SEO × AI. Search Console shows where Google ranks you but AI does not cite you, and the reverse; GA4 shows sessions and conversions that already arrive from ChatGPT, Perplexity, Gemini and Claude. Both connections are read-only and can be disconnected at any time.