Peek-Then-Pay
The Open Standard for AI Content Discovery and Use
A protocol that separates discovery from access for the first time. AI agents evaluate content before retrieving it. Publishers control what is discovered, accessed, and exchanged in return.
Why Peek-Then-Pay?
Think of it as robots.txt for AI, but enforceable and auditable. A new architecture for how AI agents and content owners interact.
⚠️ Two Broken Extremes
Unrestricted Scraping
AI agents scrape content freely, resulting in incomplete and inaccurate representation, no click, no credit, and no path to value for the publisher.
Full Blocking
Blocking AI entirely removes the publisher from discovery altogether. Content that cannot be found cannot drive value.
To date this has forced a binary choice. Neither is a viable strategy.
✅ A Third Path
Decentralized Control
Publishers signal participation through a peek.json manifest and maintain authority at their domain edge. No centralized gatekeeper controls access.
Centralized Coordination
Value exchange is coordinated across the ecosystem: licensing, attribution, referred traffic, or any model the publisher defines.
Peek-Then-Pay introduces a third path: structured discovery without full access, with permissioned use determined separately.
Protocol Flow
Automatic Discovery: AI agents encountering a Peek-Then-Pay enabled domain automatically receive a structured peek response. Discovery happens without full retrieval. Publishers remain in control from the first request.
Composable Architecture: Build vs Buy
Publishers can mix and match components based on their technical capabilities and strategic preferences.
1. Edge Enforcer
Validates JWTs and serves content with minimal latency.
2. Tools
Transform and deliver domain-specific content.
3. License Server
Centralized marketplace for AI agent discovery and payment.
How It Works
The Peek-Then-Pay Protocol
Discovery
Publishers signal participation through a peek.json manifest, enabling AI agents to discover content terms before any retrieval occurs. No prior knowledge of the domain is required.
Preview
Content requests return HTTP 203 with preview + pricing for informed decisions
License
AI agents acquire JWT licenses for specific intents (read, summarize, embed, etc.) from the license server specified in the peek.json manifest
Access
Edge enforcers validate JWTs and serve licensed content with forensic manifests. Plug-and-play tools can deliver pre-transformed data, benefiting publishers through pay-per-use models while helping AI agents skip data cleaning and transformation steps
Report
Both parties report usage asynchronously for billing and dispute resolution
💡 Key Innovation: For the first time, discovery and use are structurally separate. An AI agent can evaluate whether your content is relevant before committing to access, and you control exactly what they see during that evaluation.
⚡ Efficiency Gains: Structured peek responses reduce AI compute by over 90% compared to raw HTML retrieval. AI operators find content faster and more accurately. Publishers control what that structured response contains.
{
"version": "1.0",
"meta": {
"site_name": "TechNews Daily",
"publisher": "TechNews Corp",
"publisher_id": "01HQ2Z3Y4K5M6N7P8Q9R0S1T1X",
"domains": ["technews.com", "*.technews.com"],
"categories": ["technology", "news", "reviews"],
"last_updated": "2025-09-01"
},
"enforcement": {
"rate_limit_per_ip": 50,
"grace_period_seconds": 300,
"failover_mode": "deny",
"bypass_paths": ["/robots.txt", "/sitemap.xml"]
},
"license": {
"license_issuer": "https://api.fetchright.ai/v1/license",
"terms_url": "https://technews.com/legal/ai-terms",
"supported_intents": ["peek", "read", "summarize", "quote"],
"content_hints": {
"average_page_size_kb": 45,
"content_types": ["text/html", "application/json"],
"update_frequency": "daily",
"cache_duration": 3600
}
}
}Standard Intents
Publishers can provide pre-computed transformations, eliminating redundant AI processing and ensuring domain-specific accuracy.
🔍 Discovery
AI agents automatically discover licensing through standardized manifest files and HTTP 203 previews
⚡ Edge Enforcement
Fast, distributed license validation at CDN/edge layer with assertion-only JWTs
📊 Bilateral Reporting
Both parties track usage for accuracy, billing reconciliation, and dispute resolution
Try the Protocol
Test peek.json manifests, validate pricing schemas, and generate licenses against the live specification.
1. Validate peek.json Manifest
Test the validation of a peek.json manifest file against the official specification schema.
2. Validate Pricing Schema
Test the validation of a pricing scheme against the pricing schema specification.
3. Generate Assertion-Only License
See how the license server generates JWT licenses and DPoP proofs for content access.
These demos use the official @peekthenpay/peek-json-spec package for validation and license generation.
Ecosystem
Peek-Then-Pay
Open Source Standard
The open schema and governance framework hosted on GitHub. Community-driven development with transparent RFC process.
- • CC0-1.0 Licensed specification
- • Community governance
- • Open RFC process
- • Reference implementations
FetchRight.ai
Reference Platform
The commercial infrastructure layer for Peek-Then-Pay, handling enforcement, licensing, analytics, and value exchange at the CDN edge.
- • Secure license issuance & verification APIs
- • Token-based accounting & payment processing
- • Structured value exchange across licensing, attribution, and referred traffic models
- • Analytics for publishers and operators
The protocol is open and free to implement. Commercial services like FetchRight build on top of it, accelerating adoption without creating dependency or lock-in.
Learn More About FetchRight →Get Involved
Peek-Then-Pay is community-driven. Help shape the future of AI content discovery.
Join Discussions
Participate in discussions on GitHub Issues and RFCs. Help design new features and improvements.
View GitHub Issues →Contribute Code
Contribute adapters for Cloudflare, Fastly, and Nginx. Build tools and libraries for the ecosystem.
Browse Repositories →Spread the Word
Join the conversation on LinkedIn and help evangelize responsible AI content licensing.
Follow #PeekThenPay →Want to Implement Peek-Then-Pay?
We are looking for early adopters and implementation partners: publishers, affiliate networks, brands, AI companies, and infrastructure providers. If you want to participate in building the standard, we want to hear from you.
Contact UsGovernance & License
Open Standard
Peek-Then-Pay is an open standard released under the CC0-1.0 Universal license. Specification and governance discussions live in the PeekThenPay GitHub organization.
The patent notice covers the Peek-Then-Pay mechanism. The specification itself is CC0-1.0 Universal and free to implement without restriction. Any developer, publisher, or platform can build on the protocol without requiring a commercial relationship with FetchRight.
Principles
- •Open: Free to implement and use
- •Neutral: No vendor lock-in
- •Transparent: Public governance process
- •Collaborative: Community-driven development
Governance
- •RFC process for major changes
- •Community consensus decision making
- •Regular community calls and updates
- •Transparent roadmap and priorities
Current Version: 1.0 (Draft) •License: CC0-1.0 Universal •Repository: github.com/PeekThenPay/peek-json-spec
View Full License →