Research from our own data.
Our research comes from Radar platform audits, the AI sales agent we run on our own site, and the git history of everything we built, including what that history cannot prove.
2 open-source frameworks ( Thread-Based Engineering + Ghost Protocol, MIT licensed). 8 research pillars. 37 published studies. Every finding from our own data.
Featured Report / Updated daily
State of AI Visibility 2026: Benchmarks from 142 Domain Audits
Aggregated from 142 real Radar platform audits. Average overall Radar score, from each domain’s latest audit: 45/100.
Tool runs
2,222
Domains tested
311
Benchmarked
142
Anonymized Benchmarks / Updated Daily
AI Visibility Benchmarks by Industry
Real Radar audits, fully anonymized. No domain or brand names ever published. See the score distribution across industries and benchmark where your domain would rank against peers.
See the benchmarksTransparent Reference Index / Rolling Scans
Brand Index: 50 Curated Brands, AI Readiness Scores
Named brands across SaaS, E-commerce, Fintech, Healthcare, and Media, each scored with Radar's AI Readiness Score, a technical check of each brand's own site. See which brands pass the technical checks and which are blocked behind paywalls or anti-bot defenses.
Browse the indexSelf-Audit / Latest Run Published
We Dogfood Radar. Here’s What It Caught.
We run Radar on pixelmojo.io and publish the latest audit. The headline finding is the hallucinations it found in the AI answers it collected about our own brand. We publish the findings, including the parts that do not flatter us. If a methodology cannot survive its own scrutiny, it is not a methodology worth selling.
See what Radar caught13
AI visibility tools in one Radar audit
Free check runs 6. Full audit from $5.
515
commits behind the Radar origin story
October 2025 to April 2026, told from git.
12
Vector scoring dimensions
Production-tested on real B2B leads.
$2
Hive resolution cost
per successful resolution, no per-seat fees.
7
Thread types in TBE
Open-source kit, with human checkpoints in every type.
8
Ghost Protocol agents
MIT licensed. git clone to install.
Why proprietary data matters for AI citation
Original research produces a 67% higher AI citation rate compared to synthesized content. When an AI model needs to answer a question, it cites the primary source. If your data exists nowhere else, you become the only source it can cite.
Source: Radar methodology and internal citation tracking across 50+ domain audits.
8 research pillars
Each pillar represents a domain where we have proprietary data, production experience, or open-source frameworks we built and tested ourselves.
How we audit and fix AI visibility, told from our own commit history and Radar audit data from 50+ domains, with what the record cannot prove marked as such.
Why We Built Radar: The Commit Trail Behind Our Own AI Visibility Fixes
Apr 5, 2026
Radar's First Beta Weeks: What Our Records Show (Dated Note)
Apr 2, 2026
Best AI Visibility Tools (2026): 10 Options Compared
Apr 10, 2026
How to Track AI Citations: A Practical Guide to ChatGPT, Perplexity, Claude & Gemini
Apr 11, 2026
Your SEO Is Fine. Your AI Visibility Needs Its Own Measurement.
Apr 9, 2026
What Is AI Technical Readiness? (And Why Monitoring Alone Is Not Enough)
Apr 8, 2026
Our 10-part playbook on Generative Engine Optimization. What makes AI cite one brand over another. Tested on our own site, published with real before/after data.
GEO Playbook: Access, Content and Measurement for ChatGPT, Perplexity and Claude
Feb 17, 2026
SEO vs AEO vs GEO: From Ranking in Search to Becoming the Recommended Brand
Feb 17, 2026
How to Build a Brand That AI Search Engines Cite
Jun 12, 2025
Google Traffic Dropped 33%? What the AI-Search Shift Means
Feb 17, 2026
How We Built a Knowledge Graph That LLMs Actually Cite (With Real Data)
Feb 25, 2026
Your llms.txt Is Already Stale. Here's How to Fix It.
Feb 17, 2026
How Our AI Bot Policy Changed, November 2025 to October 2026 (Dated Note)
Mar 6, 2026
How Vector scores leads across 12 dimensions and explains each routing decision, where human review belongs, and when a simpler scorer is enough.
What happens when AI agents share context and coordinate autonomously. Enterprise deployment patterns across logistics, insurance, HR. $2/resolution economics.
Multi-Agent AI Systems Explained: When One AI Is Not Enough
Jan 2, 2026
Multi-Agent AI Platform: Build vs Buy - Pricing & TCO (2026)
Jan 5, 2026
The Dawn of Agentic AI: From Chatbots to Co-workers in 2026
Jan 10, 2026
Why AI-First Customer Service Is the New 'Press 1 for Sales' (and Why You're Losing Customers)
Oct 19, 2025
Conversation Flow Architecture [4 Design Layers]
Mar 11, 2026
We defined AX Design as a methodology. Capability mapping, trust patterns, supervision models for AI-as-coworker interfaces. The complete framework.
AX Design Explained: The 2026 Guide to Agentic Experience
Mar 6, 2026
Trust Design Patterns: How Users Learn to Rely on AI Coworkers
Mar 8, 2026
AX Metrics: How to Measure Agentic Experience Quality Beyond Task Completion
Mar 14, 2026
Agent Personality Design: Voice and Trust Framework
Mar 13, 2026
From UX to AX: What Design Looks Like When AI Becomes Your Co-Worker
Feb 17, 2026
Thread-Based Agentic Experience Engineering [TBE + AXD]
Mar 23, 2026
Our open-source governance framework. 7 thread types, Core Four fundamentals, mandatory checkpoints. Used in production on our own products, including Lakbay AI. TBE governs how work is structured. Ghost Protocol governs who does it.
Thread-Based Engineering: The Framework for Scaling AI Development
Jan 24, 2026
Thread-Based Engineering: How We Keep AI-Assisted Code Reviewable
Feb 1, 2026
Thread-Based Agentic Experience Engineering [TBE + AXD]
Mar 23, 2026
Lakbay AI: What We Built and What We Withdrew (Dated Note)
Feb 8, 2026
Context Engineering Beyond CLAUDE.md: The 5-Layer Hierarchy
Feb 14, 2026
8 named specialist agents, 9 execution patterns, dual routing. Built on Thread-Based Engineering. MIT licensed. One install command for any GitHub Copilot project.
Security flaws in 45% of AI code tests and an 86% XSS failure rate (Veracode), and 66% of developers frustrated by "almost right" AI output (Stack Overflow). We documented the data and built the governance framework we use.
84% of Developers Use AI Tools. 45% of AI Code Has Flaws.
Jan 31, 2026
Claude Code Technical Debt Mitigation: The Complete Production Guide
Feb 1, 2026
Claude Code Hooks: Corrections to Our February 2026 Guide (Dated Note)
Feb 14, 2026
The Junior Developer Extinction Problem: Why AI Technical Debt Needs Human Apprentices
Feb 25, 2026
About Pixelmojo Labs
Pixelmojo Labs is the research division of Pixelmojo, an AI product studio founded in 2024 and based in Makati, Metro Manila, Philippines. We publish original data from three sources: the Radar AI visibility platform (50+ domain audits), the Vector sales agent we run on our own site, and the commit history behind Radar.
We also maintain two open-source frameworks: Thread-Based Engineering (AI development governance) and Ghost Protocol (multi-agent orchestration for GitHub Copilot). Both are MIT licensed and available on GitHub.
All research is authored by Lloyd Pilapil (with project work for Salesforce, Parsons, and Egis) and the Pixelmojo research team.
Run the same audit on your brand.
Radar orchestrates 13 tools in staged batches on any domain. The same platform that powers our research. The first audit is free.