Why should I care what AI says about my product?

Because 85% of B2B buyers now ask AI before visiting your website. If ChatGPT says your pricing is 'expensive' or Perplexity recommends a competitor instead of you, buyers never tell you why they didn't book a demo — they just don't show up. AI buyer perception is the gap between what's true about your product and what AI tells your buyers.

What do I get that a visibility score doesn't give me?

AI visibility tools give you a score — 'you were mentioned 47 times.' That tells you nothing about deal impact. Perciva shows you the actual problems: 'AI is telling buyers your pricing starts at $79 (it's actually $29)' and 'CompetitorX replaced you on 4 comparison prompts this week.' Every alert comes with a specific fix — which page to update and what to add.

Which AI engines do you support?

We currently monitor ChatGPT, Perplexity, Gemini, and Claude — the four AI engines most frequently used by B2B buyers during software research and evaluation. Support for Copilot and other emerging AI interfaces is actively being added as buyer adoption grows.

How do I fix AI misinformation about my brand?

Perciva tells you exactly which claims are wrong and which pages to update. Common fixes include updating your pricing page with clear numbers, adding integration documentation with proof points, and publishing case studies that AI engines can cite as authoritative sources.

Do I need any technical setup?

No. Perciva is fully self-serve — you enter your brand name, add competitors, and select prompt packs. Monitoring starts automatically within minutes. There is no code to install, no integrations to configure, and no IT involvement required.

How does the free snapshot work?

Enter your brand name and work email. Within 24 hours, you'll receive a report showing how AI describes your brand on key buyer-intent prompts, including risky claims, competitor positioning, and 3 fast fixes.

Is there a free trial?

Yes. Every plan includes a 7-day free trial with full access to all features — prompt monitoring, claim extraction, competitor tracking, and alerts. No credit card required to start. You can also request a free AI Buyer Perception Snapshot with no trial needed.

How much does one wrong AI answer cost my pipeline?

If an AI engine tells a buyer your product doesn't support their key integration, and that buyer moves to a competitor without ever visiting your site, that's one lost deal. For most B2B SaaS companies, a single deal is worth $5K–$50K+ annually. One wrong claim across thousands of buyer queries adds up fast. Perciva starts at €49/mo.

How quickly do AI engines update after I fix my content?

It varies by engine. Some changes are reflected within days (especially in engines with web browsing), while others may take 2–4 weeks as models re-index sources. Perciva tracks these updates so you can confirm when your fixes have taken effect.

What if I only have 1-2 competitors?

Even with few competitors, the combination of prompt categories (pricing, integrations, security, comparisons) and AI engines (4+) creates dozens of monitoring points. A single wrong pricing claim or competitor displacement on one prompt can impact pipeline significantly.

How often is monitoring run?

All plans include weekly monitoring that automatically runs prompts across ChatGPT, Perplexity, Gemini, and Claude. You can also run on-demand checks whenever you need an immediate answer — for example, after publishing new content or launching a product update.

Can agencies use this for clients?

Absolutely. The Team/Agency plan supports up to 5 projects with white-label branded exports, making it ideal for agencies managing AI perception across multiple client brands. Each project has independent competitor tracking, prompt packs, and alert settings.

Can I see a sample report before signing up?

Yes. We've published a full sample AI Buyer Perception Report using a fictional brand so you can see exactly what you'll get — risky claims, competitor displacement alerts, citation analysis, and prioritized fixes. View it at perciva.ai/sample-report.

How is this different from SEO tools?

SEO tools optimize for Google search rankings. Perciva monitors what AI chatbots say about you on buyer-intent prompts — a completely different surface. You can rank #1 on Google and still be misrepresented in ChatGPT answers that buyers rely on.

AI Crawler Access: The robots.txt Decisions That Shape Your AI Visibility

2026-07-24 · 6 min read · By Perciva Team

Your robots.txt now controls three separate relationships with AI: whether your content trains future models, whether it appears in AI search results, and whether an AI assistant can fetch a page when a user asks about you. Each is governed by different user-agents, and blocking the wrong one silently removes you from answers your buyers are reading.

For most B2B SaaS companies the strategic answer is simple: allow AI crawlers on your public marketing and documentation pages, because being absent from AI answers is a demand-generation problem, not a content-protection win. But "allow" should be a decision, not a default you never examined. Here is the current crawler roster, what each user-agent actually does, and correct configurations for both open and selective postures.

Three Crawl Purposes, Three Decisions

Training crawlers collect content to train future models. Blocking them is a bet that protecting content outweighs being known by the models buyers use.
Search-index crawlers build the retrieval indexes behind AI search products. Blocking these directly removes you from cited, real-time answers.
User-triggered fetchers retrieve a page live because a user asked. Blocking these means when a buyer says "check Acme's pricing page," the assistant cannot.

These are independent controls. The most common misconfiguration is blocking all three when the intent was only the first.

The Crawler Roster

OpenAI

GPTBot — training. Blocking it signals your content should not train OpenAI models. It does not affect ChatGPT search visibility.
OAI-SearchBot — indexing for ChatGPT search. Block this and you disappear from ChatGPT's cited web results.
ChatGPT-User — user-triggered fetches during conversations. OpenAI notes user-initiated fetching may not behave like automated crawling.

Anthropic

ClaudeBot — training.
Claude-SearchBot — search indexing.
Claude-User — user-triggered fetches.

Anthropic documents all three as respecting robots.txt, and each requires its own directive — blocking ClaudeBot alone does not block the other two. More detail in our AI crawler glossary entry.

Perplexity

PerplexityBot — builds Perplexity's search index. Blocking it removes you from one of the most citation-forward engines B2B buyers use.
Perplexity-User — fetches pages when a user asks. Perplexity states that because these are user-initiated requests, robots.txt directives generally do not apply to them.

Google

Googlebot — classic search crawling, which also feeds AI Overviews and AI Mode.
Google-Extended — not a separate crawler but a robots.txt control token. Disallowing it opts your content out of Gemini training and Search-grounded Gemini answers. Two facts teams routinely get wrong: blocking Google-Extended does not remove you from AI Overviews (those ride on normal Google indexing), and Google has stated it does not affect Search rankings.

Others worth knowing

CCBot (Common Crawl) feeds public datasets used in many models' training. Applebot-Extended, meta-externalagent, Amazonbot, and ByteDance's Bytespider follow the same pattern of vendor-specific tokens — audit them when your policy needs to be exhaustive rather than pragmatic.

Configuration 1: The Open Posture (Recommended for SaaS)

User-agent: *
Disallow: /app/
Disallow: /api/
Disallow: /account/

Sitemap: https://acme.com/sitemap.xml

Everything public stays crawlable for everyone; only the logged-in product, API, and account surfaces are excluded. Two housekeeping notes that save real pain: keep staging and preview environments behind authentication or a blanket disallow, because a crawled staging site with placeholder pricing can leak into answers; and remember that robots.txt is per-host, so your docs subdomain needs its own correct file. If you maintain an llms.txt file, verify every URL it lists is allowed here — the two files must agree.

Configuration 2: Selective — Search Yes, Training No

Some companies (usually content-heavy ones) want AI search citations without contributing training data:

User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

User-agent: CCBot
Disallow: /

User-agent: Google-Extended
Disallow: /

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: *
Disallow: /app/

Understand the cost before copying this: models trained without your content know less about you, and what they "know" comes entirely from third parties. For a niche SaaS brand, that often means being described by your competitors' comparison pages. Being retrievable at answer time softens this, but training-time familiarity still influences which brands models reach for — being absent from it is a real trade, and for most vendors a bad one.

robots.txt Is Not the Whole Story

Your CDN may be blocking AI crawlers for you. Bot-management defaults at several major CDNs challenge or block AI user-agents regardless of robots.txt. If your visibility is mysteriously poor, check the WAF before rewriting content.
robots.txt is a request, not a wall. Reputable crawlers comply; disreputable ones do not. Enforcement requires network-level controls.
Verify with logs. Your server logs show which AI user-agents actually fetch which pages and how often. Docs and pricing pages drawing steady fetches from search-index bots is what healthy looks like.

Verify in the Logs, Not the Config

Your robots.txt states policy; your access logs state reality. A quick check on most setups:

grep -iE "gptbot|oai-searchbot|chatgpt-user|claudebot|claude-searchbot|perplexitybot" access.log | awk '{print $1, $7, $12}' | sort | uniq -c | sort -rn | head -50

Healthy looks like steady fetches from search-index bots across your docs, pricing, and comparison pages. Two anomalies worth chasing: a bot you allowed that never appears (check CDN bot management — it may be challenged before reaching your origin), and heavy fetching of pages you disallowed (either a misconfigured rule or a spoofed user-agent; major vendors publish IP ranges so you can verify authenticity before accusing anyone of noncompliance).

Frequently Asked Questions

If we block GPTBot, do we disappear from ChatGPT?

No — and this is the most consequential nuance on the page. Blocking GPTBot only opts you out of future training. ChatGPT can still cite you via search (OAI-SearchBot) and still "knows" whatever earlier training absorbed. Conversely, allowing GPTBot does not get you into ChatGPT search — that is OAI-SearchBot's job. Match the user-agent to the outcome you actually care about.

Should we block Bytespider and other aggressive crawlers?

ByteDance's Bytespider has drawn persistent complaints about crawl aggression, and crawlers with poor compliance reputations make robots.txt directives unreliable as enforcement. If you decide to block a crawler you do not trust, do it at the WAF or network level and treat the robots.txt line as documentation of intent.

Does any of this affect our Google rankings?

Blocking AI-specific tokens like Google-Extended does not affect classic Search rankings — Google has said so explicitly. Blocking Googlebot itself, of course, is a different and much larger decision.

The Decision Checklist

Decide the three postures explicitly: training, search indexing, live fetch.
Never block search-index bots (OAI-SearchBot, PerplexityBot, Claude-SearchBot) on pages you want cited — see how to get cited by Perplexity.
Keep pricing, security, docs, and comparison pages crawlable — they answer the buyer questions with money attached.
Audit CDN and WAF rules against your robots.txt intent.
Re-run the audit quarterly; vendors add user-agents without ceremony.

Access Is the Prerequisite, Not the Prize

Correct crawler access makes you quotable; it does not make you quoted. Once the gates are open, the question becomes which sources engines actually cite when buyers ask about your category — and whether any of them are yours. That is citation territory: measure it, then close the gaps. Perciva tracks which pages AI engines cite on real buyer questions, so you can see whether opening the gates changed the answers.

← Back to Blog

Perciva — AI Buyer Perception Monitor for B2B SaaS

Perciva is an AI Buyer Perception Monitor that tracks what ChatGPT, Perplexity, Gemini, and Claude say about your B2B SaaS product on buyer-intent prompts. It detects incorrect claims, competitor displacement, and citation loss before they cost you pipeline.

How It Works

Configure monitoring — Add your brand, competitors, and select prompt packs.
Monitor buyer prompts — Perciva runs prompts across major AI engines and captures full answers.
Detect changes & risks — See what changed in claims, recommendations, and citations.
Alerts & action hints — Get email alerts on critical changes with clear action hints.

Key Features

Answer Capture — Full AI answer snapshots from ChatGPT, Perplexity, Gemini
Answer Diffs — Side-by-side diff showing exactly what changed
Claim Extraction — Auto extract claims about pricing, features, positioning
Risk Flags — Alerts for incorrect claims, weakened positioning, displacement
Competitor Tracking — Track when competitors gain or lose visibility
Citation Monitoring — See which URLs AI cites
Email Alerts — Notifications on critical changes
Action Hints — Each alert includes a recommended fix

Pricing

Starter: €49/mo · Growth: €129/mo · Team: €299/mo — All plans include a 7-day free trial.

FAQ

What is AI buyer perception?: AI buyer perception is how AI engines like ChatGPT, Perplexity, Gemini, and Claude describe and position your product when potential buyers ask comparison, pricing, or evaluation questions.
How is Perciva different from AI visibility tools?: Most AI visibility tools give you a score or mention count. Perciva monitors the actual buyer-intent prompts that drive purchasing decisions and tells you exactly what AI is saying about your brand.
Which AI engines do you support?: We monitor ChatGPT, Perplexity, Gemini, and Claude.
How do I fix AI misinformation about my brand?: Perciva tells you exactly which claims are wrong and which pages to update. Common fixes include updating your pricing page, adding integration documentation, and publishing case studies.
Do I need any technical setup?: No. Perciva is fully self-serve. No code, no integrations, no IT involvement.
Who is this product for?: B2B SaaS companies with 10–200 employees. Ideal for Heads of Marketing, SEO Leads, Product Marketing, and Growth teams.
What is competitor displacement in AI answers?: Competitor displacement occurs when an AI engine starts recommending a competitor instead of your product on buyer-intent prompts.
What is GEO (Generative Engine Optimization)?: GEO is the practice of optimizing your web content so that AI engines cite, reference, and accurately represent your brand in their responses.
How does the free snapshot work?: Enter your brand, website, a main competitor, and your category. Within 24 hours, you receive a report showing how AI describes your brand on key buyer-intent prompts — including risky claims, competitor positioning, and 3 fast fixes.
Is there a free trial?: Yes. Every plan includes a 7-day free trial with full access to all features. No credit card required. You can also request a free AI Buyer Perception Snapshot with no trial needed.
How often is monitoring run?: All plans include weekly monitoring across ChatGPT, Perplexity, Gemini, and Claude. On-demand checks are always available.
Can agencies use this for clients?: Absolutely. The Team/Agency plan supports up to 5 projects with white-label branded exports — ideal for agencies managing AI perception across multiple client brands.
How is this different from SEO tools?: SEO tools optimize for Google search rankings. Perciva monitors what AI chatbots say about you on buyer-intent prompts — you can rank #1 on Google and still be misrepresented in ChatGPT.

Who Is Perciva For?

Perciva is built for B2B SaaS companies with 10–200 employees who have clear competitors and a category where buyers research and compare using AI. Ideal users are Heads of Marketing, SEO Leads, Product Marketing Managers, and Growth teams.

Resources

Contact: hello@perciva.ai · perciva.ai