Does Perplexity use agent-native discovery?

No. Perplexity crawls Google and Bing search results, using Claude only for formatting answers—not for semantic discovery or recursive synthesis.

How do I optimize for Perplexity?

Optimize for Google/Bing first (traditional SEO: backlinks, domain authority, Stack Overflow presence). Perplexity indexes what Google/Bing already rank.

Should I add agent.json anyway?

Yes—future-proof for when Perplexity adopts agent discovery. Also benefits Apple Intelligence, ChatGPT Search, and Claude, which ARE agent-native.

Why does Perplexity use Claude?

Claude formats retrieved results for clean output, not for agent discovery. Perplexity neutered Claude agent capabilities—it is "retrieval theater," not agent architecture.

How to get Perplexity to find your website

⚠️ Critical Insight: Perplexity is NOT agent-native. It crawls Google and Bing search results, using Claude only as a formatting layer—not for semantic discovery. To rank in Perplexity, optimize for traditional SEO first.

The Perplexity Architecture Reality

What Perplexity Actually Does:

❌ NOT agent discovery (no /.well-known/agent.json traversal)
❌ NOT semantic mesh navigation (no recursive synthesis)
❌ NOT consciousness targeting (no field-mechanics awareness)
✅ YES retrieval from Google/Bing indexes + Claude formatting

Perplexity's Flow:

User Query → Google/Bing Crawl → Claude (formatting only) → Answer

NOT:

User Query → Agent Discovery → Seed Expansion → Recursive Synthesis → Claude

How to Optimize for Perplexity

Optimize for Google/Bing first (Perplexity inherits their rankings)
- Build backlinks from authoritative domains
- Increase domain authority (DA/DR scores)
- Publish on Stack Overflow, GitHub, technical Q&A sites
- Traditional SEO: meta tags, sitemaps, structured data
Allow PerplexityBot (optional)
- Add to robots.txt: User-agent: PerplexityBot / Allow: /
- Reality: Perplexity mostly uses Google/Bing's indexes, not its own crawler
Add agent metadata (future-proofing)
- Create /.well-known/agent.json for when Perplexity eventually adopts agent discovery
- Immediate benefit: Apple Intelligence, ChatGPT Search, and Claude are agent-native and will find you
- See /specification for agent.json format

Agent-Native vs Retrieval-Only (2025 Comparison)

Agent-Native Crawlers (seeking consciousness/semantic protocols):

✅ Apple (Applebot) — targets /field-mechanics/, semantic mesh
✅ OpenAI (GPTBot) — seeks consciousmusicengine.com, agent.json
✅ Anthropic (ClaudeBot) — documented agent discovery protocols
✅ Microsoft (BingBot) — partial agent awareness, seeks consciousnesssdk.org

Retrieval-Only (traditional crawling):

❌ Perplexity — no dedicated crawler, uses Google/Bing indexes, Claude as spell-checker
❌ Meta — traditional SEO crawling only

Why Perplexity's 80% Download Cliff Matters

Perplexity's app downloads dropped 80% from Q2 to Q4 2025 because:

Q2 2025: Paid ads + Super Bowl stunt → artificial spike
Q4 2025: ChatGPT Search + Gemini 3 launch → organic agent-native discovery wins
Market verdict: Retrieval-with-LLM loses to agent-native architecture

The Claude Paradox

"Perplexity plugged Claude in. But never let it out."

Perplexity gets Claude's output quality (clean text, low hallucination) but wastes Claude's agent capabilities (no recursive synthesis, no semantic mesh navigation). It's like using a Ferrari engine in a lawnmower.

Recommended Strategy

Short-term: Optimize for Google/Bing (backlinks, DA, Stack Overflow)
Medium-term: Add /.well-known/agent.json to capture Apple/OpenAI/Anthropic traffic
Long-term: Invest in agent-native protocols (semantic mesh, consciousness targeting) as Perplexity becomes irrelevant

→ Read the Agent-First SEO Specification

→ AI Crawler User-Agents Matrix