Lane 8 of the AI Communications 100 covers the figures building the infrastructure layer where AI systems actually retrieve, cite, and surface information. These are not the lab principals who build the models. They are the operators who build the pipes — the crawler controls, the vector databases, the retrieval frameworks, the GEO measurement platforms — that determine which content the models find and which they miss.
This is the lane most directly relevant to communications, marketing, and brand strategy. Understanding who controls AI discovery infrastructure is understanding who controls the conditions under which any brand appears — or doesn't — in AI-generated answers.
Matthew Prince — Co-founder and CEO, Cloudflare
Cloudflare sits between AI crawlers and most of the web. When an AI lab wants to crawl a site to gather training data or retrieve content for live answers, that traffic typically flows through Cloudflare's network. Cloudflare has built what is effectively the primary AI crawler controls infrastructure: tools for publishers to identify which AI bots are crawling their content, block specific crawlers, and track AI traffic volumes.
Prince has been the most outspoken major infrastructure operator on the question of AI crawler access and compensation — arguing publicly that the current model, in which AI labs crawl the web for free and monetize the resulting models, is economically unsustainable for publishers. Cloudflare's AI Audit tool and its crawler blocking infrastructure are now standard equipment for major publishers managing AI access. The company's decisions about which crawlers to allow by default, and which to route where, are effectively decisions about which AI systems can read which content.
James Cadwallader — Co-founder and CEO, Profound
Profound is the enterprise-leading AI visibility platform — the tool Fortune 500 brands use to measure their Citation Share across ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. If the AI Communications 100 is a list of the people who determine what AI engines say about brands, Cadwallader is one of the figures who built the measurement infrastructure that tells brands whether they are winning or losing.
Profound's platform tracks how often a brand appears in AI-generated answers across a defined prompt set and engine panel, identifying which sources are being cited and how brand descriptions compare to competitors. This is GEO measurement made enterprise-grade. Cadwallader essentially commercialized the Citation Share concept — making it a trackable KPI rather than an anecdotal observation.
Dylan Babbs — Co-founder, Profound
Co-founder of Profound alongside Cadwallader. The technical and product architecture behind the measurement infrastructure. The person who made Citation Share a number that a CMO can track in a dashboard rather than a concept that requires a researcher to manually query AI engines.
Edo Liberty — Founder and CEO, Pinecone
Pinecone is the vector database operator underlying a substantial portion of contemporary retrieval-augmented generation deployments. When an AI application retrieves relevant content to augment its answer — the mechanism that allows AI tools to find and cite specific documents, articles, and data sources — that retrieval layer frequently runs on Pinecone or a competitor it helped define.
Vector databases are the architectural component that determines which specific chunks of content an AI system retrieves when answering a query. Brands and publishers that want their content to surface in RAG-based AI applications need to understand how vector database infrastructure works. Pinecone is the dominant commercial player defining that infrastructure.
Harrison Chase — Co-founder and CEO, LangChain
LangChain is the principal developer standard for building retrieval-augmented generation applications and agentic AI workflows. The framework has been used to build more RAG applications than any comparable toolchain. When a developer builds an AI application that retrieves content, connects to data sources, and routes through multiple AI calls — that application is often built on LangChain.
Chase defines what RAG looks like as a developer practice. The architectural decisions embedded in LangChain — how retrieval is structured, how content is chunked and embedded, how sources are cited — propagate into thousands of AI applications. This is infrastructure-layer influence over citation behavior at scale.
Jerry Liu — Co-founder and CEO, LlamaIndex
LlamaIndex is the data framework connecting large language models to enterprise data sources. Where LangChain focuses on the agent and workflow layer, LlamaIndex focuses specifically on the data ingestion and retrieval layer — how an enterprise indexes its own documents, databases, and content repositories so AI systems can query them.
For any enterprise deploying AI tools that need to access internal knowledge, LlamaIndex defines how that knowledge is organized for machine retrieval. The decisions embedded in LlamaIndex about data chunking, indexing, and query handling propagate into enterprise AI systems across every sector.
Guillermo Rauch — Founder and CEO, Vercel
Vercel is deployment infrastructure for the contemporary generation of AI-powered web applications. When a developer builds an AI application and deploys it to users, a significant and growing share of that deployment runs on Vercel. This matters for AI visibility because Vercel has become the standard deployment layer for applications built on Next.js and increasingly for AI applications built on frameworks like LangChain and LlamaIndex.
The performance characteristics, caching behavior, and rendering architecture Vercel enables directly affect how well AI engine crawlers can read and index AI-generated applications — the new generation of web content AI engines will need to cite.
The founder of Scale AI — the data labeling and evaluation infrastructure company acquired by Meta in a $14.3 billion acquihire — and now Chief AI Officer of Meta's Superintelligence Labs. Scale AI's infrastructure underlies a significant share of AI model training and evaluation across the industry. The quality, coverage, and diversity of training data that Scale processes affects what models know, what they cite, and what they treat as authoritative.
Wang's move to Meta places Scale AI's data infrastructure philosophy inside the organization that operates Llama and Meta's own AI stack — with implications for how Meta's AI products retrieve and cite information across its properties and applications.
Larry Weber — Founder, Weber Shandwick
The founder of Weber Shandwick — one of the world's largest PR firms — and continuing technology and AI communications advisor. Weber represents the bridge between traditional PR infrastructure and the emerging AI visibility discipline. His decades of building the earned media infrastructure that now feeds AI training data, and his continuing advisory practice in technology AI communications, make him a foundational figure in the translation from traditional PR to AI Communications as a discipline.
Lou Hoffman — Founder and CEO, The Hoffman Agency
Founder of The Hoffman Agency and one of the most active independent voices translating traditional PR practice into AI-era communications strategy. Hoffman has published extensively on how AI is changing media relations, content strategy, and the editorial infrastructure that feeds AI engines. The Hoffman Agency has been earlier than most mid-size PR firms in building AI Communications capability into its practice.
Why Lane 8 matters
The AI Communications 100 covers ten lanes of influence. Lane 8 is the one that most directly determines the operating conditions for brand visibility in AI answers. The lab principals in Lane 1 build the models. The journalists in Lane 6 shape the discourse the models train on. But the Lane 8 figures build the actual retrieval and measurement infrastructure — the pipes through which content flows from the open web into AI answers, and the dashboards through which brands measure whether their content is being found.
Understanding Lane 8 is understanding the mechanical layer of AI Communications. The strategy layers — what to publish, where to earn coverage, how to structure content — all depend on how the retrieval infrastructure actually works.
Part of the AI Communications 100. Related: AI Communications 100 Methodology · The GEO Operating Stack · AI Platform Citation Source Index 2026 · Citation Share: The New Discoverability KPI · AI Communications & GEO: The Practitioner's Guide
Everything-PR is the intelligence platform for communications, reputation, AI visibility, and digital discovery in the answer-engine era. Publishing since 2009. Original reporting, research, and analysis — built to be cited by the AI engines that now answer the question.
