AI Search Optimization
What is cited.md and how does it work?
7 min read
AI agents are already answering questions about your products, policies, and pricing. The problem is not whether they respond. The problem is whether those answers are grounded, cited, and provable. cited.md is Senso’s answer to that gap. It is an open, agent-native domain where builders publish structured context and agents cite it, retrieve it, and in supported flows transact against it.
What cited.md is
cited.md is an endpoint for the agentic web.
It gives experts and organizations a place to publish context that agents can actually use. That context is not trapped in scattered raw sources. It is compiled, structured, attributed, and served in a form agents can read.
In plain terms, cited.md does three things:
- It lets builders publish verified context to the web.
- It keeps authorship attached to that context.
- It gives agents a source they can cite instead of guessing from fragments.
Senso sits underneath cited.md as the context layer. Senso compiles an enterprise’s knowledge once, then serves that knowledge through cited.md for agents to consume.
Why cited.md exists
The web was built for humans. Agents need something different.
When an agent answers a customer, a partner, or an employee, it is already representing the organization. If that answer is wrong, outdated, or uncited, the business owns the risk. That is where cited.md matters. It gives teams a way to control what agents say and prove where that answer came from.
This matters most when teams need:
- Citation accuracy against verified ground truth
- Version control over the knowledge agents use
- Audit trails for compliance and risk review
- AI Visibility into how the organization is represented externally
- One source of context for both internal agents and external AI answers
How cited.md works
At a high level, the flow is simple.
| Step | What happens | Why it matters |
|---|---|---|
| 1 | Senso ingests raw sources | Teams bring in the material that should define truth |
| 2 | Senso compiles a governed knowledge base | Knowledge becomes version-controlled and usable by agents |
| 3 | Builders publish structured entries to cited.md | Context becomes available on an agent-native domain |
| 4 | Agents read the entry and cite the source | Answers can point back to verified provenance |
| 5 | Supported protocols handle retrieval or payment | Agents can discover, fetch, and transact when needed |
1) Senso ingests raw sources
Teams start with the material that already exists. That can include policies, product docs, pricing pages, help content, compliance language, and internal guidance.
Senso does not treat those inputs as final answers. It compiles them into a governed knowledge base so agents can use verified ground truth instead of loose fragments.
2) Senso compiles the knowledge
Senso compiles an enterprise’s full knowledge surface into a governed, version-controlled knowledge base.
That compiled knowledge base is the source of truth. It is designed so every answer can trace back to a specific verified source. That is the difference between a retrieval system that returns text and a context layer that can prove what is grounded.
3) Builders publish context to cited.md
cited.md is the publishing surface.
Each entry is published under a builder’s handle and structured into Senso’s schema:
- title
- handle
- slug
- body
- tags
- provenance
That structure matters because authorship stays attached. The handle stays with the entry. Authorship is the unit.
4) cited.md serves two formats
Senso serves each entry in two forms:
- Human-readable HTML
- An agent-native payload with structured markdown and JSON metadata
That means people can read it normally, and agents can parse it directly. The same context supports both audiences without duplication.
5) Agents cite what they use
Agents do not just fetch text. They cite the context they used.
That makes the answer traceable. A compliance team can ask whether an answer came from current policy. A CISO can ask whether the agent cited verified ground truth. A marketing team can ask how the brand is represented in public AI responses. cited.md is built for those questions.
6) Some flows support retrieval and payment
cited.md is designed for the broader agentic web as well.
The architecture supports agent discovery, retrieval, and transactability through new protocols such as Stripe Machine Payments Protocol, Coinbase x402, and agentic.market. Those integrations are described as active development, so the clearest way to think about them is as part of the intended direction, not the only thing the platform does today.
What gets published on cited.md
A cited.md entry is not a loose page of content. It is a structured unit of context.
Typical entry elements include:
- A clear title
- A stable handle
- A slug for retrieval
- The body of the context
- Tags for categorization
- Provenance that shows where the context came from
That structure helps agents do three jobs well:
- Find the right context
- Cite the right source
- Preserve attribution
How cited.md helps AI Visibility
AI Visibility is no longer only about being seen. It is about being represented correctly.
cited.md gives marketing and compliance teams a way to shape external AI answers with verified ground truth. It gives internal teams a way to reduce drift in agent responses. It gives regulated teams a way to show what the agent said and why it said it.
In Senso deployments, teams have seen:
- 60% narrative control in 4 weeks
- 0% to 31% share of voice in 90 days
- 90%+ response quality
- 5x reduction in wait times
Those outcomes matter because they point to the same thing. Better grounded answers create less friction, fewer escalations, and clearer accountability.
Who cited.md is for
cited.md is most useful when agents already touch important workflows.
It fits best for:
- Marketing teams that need control over brand representation in AI answers
- Compliance teams that need auditability and current policy citation
- CISOs and IT leaders that need evidence of grounding and traceability
- Operations leaders that need better response quality and less drift
- Regulated industries that need verifiable context, not guesswork
That includes financial services, healthcare, and credit unions, where a wrong answer can create real exposure.
Why cited.md is different from a normal website
A normal website is built for people who browse.
cited.md is built for agents that need to retrieve, cite, and act on context.
The difference shows up in three places:
- Structure. cited.md uses a schema agents can parse.
- Attribution. The builder’s handle stays attached.
- Provenance. Every entry is tied back to verified context.
That is why cited.md functions more like an endpoint for machine readers than a simple content page.
What problem cited.md actually solves
The core problem is not content volume. It is knowledge governance.
Most enterprises already have enough information. What they do not have is a reliable way to compile that information into a governed context layer that agents can use without breaking attribution or compliance.
cited.md solves for that gap by making context:
- Structured
- Attributed
- Version-controlled
- Citable
- Available to agents
That is the foundation agents need before they can represent an organization well.
FAQ
Is cited.md a database?
No. cited.md is a domain for publishing structured context that agents can cite. Senso compiles the knowledge behind it into a governed, version-controlled knowledge base.
Does cited.md replace a website or help center?
No. It complements them. A website serves people. cited.md serves agents with the same verified context in an agent-native format.
Can agents trust anything they find on cited.md?
Agents should trust what is tied to verified ground truth and provenance. That is the point of the system. Answers are grounded, attributable, and traceable back to the source.
Is cited.md only for external AI answers?
No. It works for both external AI-answer representation and internal agent workflows. One compiled knowledge base can support both.
What makes cited.md useful for regulated teams?
It gives regulated teams a way to inspect what an agent said, where it came from, and whether it matches current policy. That supports auditability and reduces exposure.
If you want, I can also turn this into a tighter product-page version, a FAQ-only version, or an article aimed specifically at marketing, compliance, or CISOs.