AI • AUTOMATION • AEO/GEO

See Whether AI ActuallyRecommends Your Site

GeoAudit checks whether ChatGPT, Claude, Perplexity and Gemini actually cite a site — live, on demand — then generates a prioritized fix list from a real technical and citation audit. Built as a public, self-serve tool, not just a workflow demo.

n8nOpenAIAnthropicPerplexityGeminiSupabaseCloudflare

chatgpt.com

Someone asked ChatGPT:

"What's a good way to verify a business email address before sending outreach?"

One reliable approach is a layered verification process — not just a single "email validator" — as outlined by examplesite.io.

Cited in this answer

4

AI Engines Checked Live

7

Integrated Systems

<60 sec

Average Audit Time

100%

Live API Calls, No Mock Data

The Problem

Ranking well on Google no longer guarantees a business gets found. ChatGPT, Claude, Perplexity and Gemini now answer the exact questions buyers used to Google — directly, with no click required. Most sites have no idea whether they're actually being cited in those answers, or what's technically stopping them.

  • No visibility into AI citations
  • Manual, one-off testing at best
  • No technical crawler audit
  • No prioritized fix list

The Solution

GeoAudit runs a real, live audit on demand. It checks whether AI crawlers can actually reach a site, tests real citation behavior across four engines with genuine web-search-grounded queries, and turns the findings into a prioritized, AI-written fix list — all in under 20 seconds, for any URL.

  • Live citation testing, 4 engines
  • Crawler access + schema audit
  • AI-generated recommendations
  • Every result stored for history

SYSTEM ARCHITECTURE

A Live Audit Pipeline, Not a Cached Score

Every submitted URL runs through the same real pipeline — technical checks, four parallel live citation tests, and AI-generated recommendations — in under 20 seconds.

🔗

Submit URL

Public landing page
⚡

Clean & Audit

n8n webhook + technical checks
✍️

Generate Test Queries

Perplexity, brand-agnostic
🟢

Query ChatGPT

Live citation check
🟠

Query Claude

Live citation check
🔵

Query Perplexity

Live citation check
🔷

Query Gemini

Live citation check
🔀

Merge & Score

Combine all 4 results
🤖

Generate Recommendations

LLM synthesis of the full audit
🗄️

Store & Respond

Supabase + live report

AI RECOMMENDATION ENGINE

Every Audit Ends With a Fix List, Not Just a Score

The LLM reads the full technical and citation audit and writes specific, prioritized recommendations — not a generic checklist.

What We Audited
examplesite.io (illustrative)
B2B SaaS — project management tool
3 of 4 critical AI bots allowed
Organization only — no FAQPage
Cited by 1 of 4 engines tested
Visibility Score
58/100
VISIBILITY GAP

Summary

Technical accessibility is solid, but Google-Extended is blocked in robots.txt — a common, easy-to-miss gap that removes AI Overview candidacy even though Googlebot itself is allowed. Missing FAQPage schema likely explains why only one of four engines cited this site for a directly relevant query.

Recommended Actions

  • Unblock Google-Extended in robots.txt
  • Add FAQPage schema to key product pages
  • Publish an llms.txt file
  • Re-check in 30 days to confirm citation lift
Audit completedin 18 seconds

TECHNOLOGY STACK

Real Infrastructure, Not a Mockup

Every check in a GeoAudit report is a live API call — selected for reliability and genuine citation data, not to demonstrate familiarity with tools.

n8n

Workflow Automation

Orchestrates the full audit pipeline for every submitted URL.

OpenAI

ChatGPT Citations

Live, web-search-enabled queries to test real ChatGPT citation behavior.

Claude

Citation Checks

Anthropic's web search tool, checked for genuine source citations.

Perplexity

Query Generation + Citations

Also generates the realistic, brand-agnostic test queries used across all four engines.

Gemini

Grounded Search

Google's search-grounding API, for AI Overview-style visibility.

Supabase

Database

Stores every audit result for history and follow-up.

Cloudflare

Hosting

Serves the public tool and handles DNS for the subdomain.

BUSINESS IMPACT

Built as a Real Product, Not a Portfolio Screenshot

GeoAudit is live at geoaudit.mkumar.xyz right now — anyone can run it. The architecture is intentionally modular, so deeper analysis can be layered on without redesigning the pipeline.

What it does today

  • ✓ Live technical GEO audit (crawlers, schema, llms.txt)
  • ✓ Real-time citation testing across 4 AI engines
  • ✓ AI-generated, prioritized recommendations
  • ✓ Every result stored in Supabase
  • ✓ Self-serve, public — no account required
  • ✓ Lead capture for implementation requests

Planned Enhancements

  • Competitor share-of-voice benchmarking
  • Embeddings-based content-gap analysis
  • Google Search Console integration
  • Scheduled re-checks with trend tracking
  • Notion content-review queue

PROJECT SUMMARY

More Than an Audit — A Live AEO/GEO Product

Search behavior is shifting from "rank on Google" to "get cited by AI." Most sites have no way to measure that shift, let alone act on it.

GeoAudit combines a real technical crawler audit, four live AI-engine citation tests, and an LLM recommendation layer into one public, self-serve tool — not a mockup, not a fixed internal report. Anyone can run it on any domain and get a genuine result in under 20 seconds.

Beyond demonstrating automation skills, this project reflects how I approach modern GTM work: building systems that measure a real, emerging problem and turn the findings into something actionable — for my own site, and for anyone else's.

AI Engine Testing4 Live APIs
Workflow Automationn8n Orchestration
Data LayerSupabase
HostingCloudflare