Is your site findable by AI agents?
Pharos makes businesses structurally discoverable and invokable by ChatGPT, Claude, Perplexity, and every agent built on MCP. We implement the technical layer — llms.txt, MCP server, structured capability data, agent-parsable content — and monitor the results.
Half of B2B buyers now start research in AI chatbots
YoY growth in AI-sourced sessions
of AI traffic arrives with no referrer (invisible in GA4)
of B2B buyers begin research in AI, not search
The agentic discovery shift is well underway. Most sites aren't structured to be found, understood, or invoked by agents. SEO agencies are rebadging old content playbooks. The technical layer is missing — and that's what we build.
Six dimensions of agent discoverability
Our Agent Discoverability Score evaluates the full technical stack. Each dimension maps to a specific service deliverable.
llms.txt Quality
15%Machine-readable sitemap telling agents what your site offers, and where.
MCP Server Discoverability
20%Your agent-invocable endpoint, discoverable via .well-known/mcp.json.
OpenAPI / API Catalog
10%Structured API description for service-based businesses.
Structured Capability Data
20%JSON-LD schema: Organization, Service, Offer, FAQPage.
Agent-Parsable Content
15%Pages that render cleanly without JavaScript; pricing in text, not images.
Citation Visibility & Monitoring
20%Live audit of where you're cited across ChatGPT, Claude, Perplexity, Gemini.
From first check to full deployment
Free to start, instant delivery on core tiers. Humans only when the work genuinely requires it.
Score
FreeURL-input scan, public grade, emailed gap report.
Audit
InstantRich PDF in 60 seconds. Live citation data, competitor comparison.
Implementation
<24hAutomated build. Full stack deployed in <24h.
Custom
Call requiredBespoke builds. Complex APIs, multi-region, custom tools.
Pharos is its own first reference implementation
Every layer we sell is live on our own brand. You can inspect the dogfood directly — no login, no sales call.
- llms.txt —
https://this-site/llms.txt - MCP Server —
https://pharos-mcp.pharos-dev.workers.dev/mcp— six tools, includingcheck_llms_txtyou can call against any site - Server Card —
https://pharos-mcp.pharos-dev.workers.dev/.well-known/mcp.json— SEP-1960 discovery metadata
When the Score tool goes live, this page will self-report its own score here.
Built for
SaaS companies
SaaS companies that want to show up when ChatGPT, Claude, and Perplexity answer questions in their category. You know the problem isn't content — it's the technical layer underneath. We build it.
Marketing & SEO agencies
Marketing and SEO agencies whose clients are asking about AI search. White-label our methodology: you bring the relationship, we bring the technical delivery. Same infrastructure, your brand.
Common questions
- What is agent discoverability?
- When AI agents like ChatGPT, Claude, or Perplexity answer questions on behalf of users, they don't browse like humans do — they fetch structured signals: llms.txt files, MCP servers, OpenAPI specs, JSON-LD schema. If your site doesn't expose these signals correctly, agents can't find you, recommend you, or invoke your services. Agent discoverability is the technical layer that makes a business legible to agents.
- How is this different from SEO?
- Traditional SEO optimizes for human search behavior on Google. Agent discoverability optimizes for programmatic consumption by language models — an entirely different surface. Some techniques overlap (clean URLs, structured data), but the core deliverables — llms.txt, MCP servers, agent-parsable content — are AEO-specific. Most SEO agencies don't do this work yet.
- Do I need a developer to use Pharos?
- No. The free Score tool and the $79 Audit are entirely self-serve — paste your URL, get results. The $1,299 automated Implementation also doesn't require your developer to write any code; we email a Git-applicable patch file that your developer applies with `git am` in about five minutes and reviews via your normal PR workflow. We never need access to your repo or infrastructure. Only Custom builds (from $4,999) involve back-and-forth with your engineering team.
- What's the difference between the free Score and the paid Audit?
- The free Score gives you a public 0–100 grade across multiple dimensions of agent discoverability with summary findings. The $79 Audit goes deeper: live citation audit across the major AI engines, competitor comparison, implementation effort estimates per gap, and a machine-readable JSON export. Audit is delivered in about 60 seconds.
- Will my customers see anything change on my site?
- No. All the changes Pharos makes are invisible to human visitors — they live in /llms.txt, /.well-known/ files, JSON-LD schema, and content reformatted for agent parseability. Your existing design, copy, and brand are untouched.
- How do I know agents are actually using the optimizations?
- Both subscription tiers include a 6-section monthly report covering MCP invocations, agent-fetched pages, AI-referrer click-throughs, citation share across ChatGPT/Claude/Perplexity/Gemini, and conversion attribution. About 70% of AI-driven traffic doesn't carry a referrer in 2026 — measuring it well requires the layered measurement stack we operate.
- What if AI search loses momentum and this becomes irrelevant?
- Possible but unlikely on the 1–3 year horizon — Shopify reports AI-attributed orders grew 11x between Jan 2025 and Mar 2026. Even if growth slows, the AEO infrastructure we install (llms.txt, JSON-LD, MCP) doesn't break or harm your traditional SEO; it's net-positive in any scenario. Nothing we deploy is reversible work.