Machine Relations: Why PR Is Becoming Citation Engineering

Answer Capsule

Machine Relations is an emerging communications discipline focused on improving the external information environment that AI systems use when evaluating a brand, product, service, or category. Instead of asking only where a company can earn coverage or backlinks, Machine Relations asks which sources actually influence machine-generated conclusions, what those sources say, which evidence they rely on, and how a brand can legitimately improve its representation within that evidence network.

In simple terms:

Traditional PR manages relationships with people and publications. Machine Relations manages the evidence environment that machines read through those people and publications.

Public relations has always been about influence.

For decades, the operating model was straightforward.

Identify the journalists, editors, analysts, creators, and publications capable of shaping public opinion.

Build relationships with them.

Give them useful information.

Earn coverage.

That coverage creates:

• awareness,
• credibility,
• reputation,
• referral traffic,
• search visibility,
• and, eventually, sales.

Digital PR added another layer.

Now the publication itself became valuable because a link could improve search visibility.

The question shifted from:

“Who can influence our market?”

to:

“Who can influence our market and provide an authoritative backlink?”

AI search is forcing the question to evolve again.

The next question is:

“Which sources are actually shaping what machines believe about our market?”

That is the territory I call Machine Relations.

What Is Machine Relations?

Machine Relations is the practice of identifying, understanding, and improving the external evidence environment that influences AI-generated answers.

It combines elements of:

• public relations,
• digital PR,
• reputation management,
• entity optimization,
• competitive intelligence,
• citation analysis,
• source provenance,
• LLM optimization,
• and information architecture.

The objective is not to manipulate an AI model directly.

The objective is to understand the public information ecosystem from which AI systems form conclusions.

That distinction is critical.

Machine Relations does not begin with:

“How do we make ChatGPT say something?”

It begins with:

“Why is ChatGPT saying it?”

AI Systems Do Not Form Opinions in a Vacuum

When an AI system recommends one company over another, marketers sometimes treat the answer like magic.

It is not magic.

The exact internal process varies by platform, and no outside agency has access to every proprietary ranking and retrieval mechanism.

But modern generative systems increasingly combine language models with external information retrieval.

The foundational research on Retrieval-Augmented Generation, published by Patrick Lewis and colleagues, demonstrated how language generation can be combined with retrieved external documents for knowledge-intensive tasks.[1]

Commercial AI-search systems are far more complex than that original architecture.

But the strategic principle remains:

The information available to the machine matters.

If the public evidence surrounding Company A repeatedly says:

• more reliable,
• better tested,
• easier to use,
• better for a specific buyer,
• better customer service,

while Company B's strongest claims exist almost entirely on Company B's own website, we should not be surprised if machines prefer Company A.

The machines are operating inside an evidence environment.

Machine Relations is about understanding that environment.

Traditional PR Targets Audiences. Machine Relations Maps Evidence.

A traditional PR campaign might begin with:

Which publications reach our customers?

A digital PR campaign might add:

Which of those publications also have strong SEO authority?

Machine Relations adds another question:

Which of those publications repeatedly appear inside the evidence networks associated with AI answers?

That creates an entirely different prioritization model.

A niche publication may have:

• modest traffic,
• modest traditional SEO metrics,
• a narrow audience,

yet repeatedly appear across high-intent AI recommendations.

Another publisher may have:

• enormous traffic,
• a powerful domain,
• millions of readers,

but very little observable influence over a particular category of AI answers.

Both can be valuable.

But they are valuable for different reasons.

Machine Relations attempts to identify that difference.

The New PR Brief

Imagine a CMO asking:

“Why does AI keep recommending our competitor?”

The old marketing stack may produce answers like:

• They have more backlinks.
• They rank for more keywords.
• They have more press coverage.
• They spend more on paid media.
• They have higher social engagement.

All of those may be true.

None necessarily explains the AI recommendation.

A Machine Relations brief should instead investigate:

1. Which AI systems recommend the competitor?
2. For which query classes?
3. Which claims are repeatedly associated with the competitor?
4. Which sources support those claims?
5. Which of those sources are repeatedly surfaced across models?
6. Which sources appear to carry disproportionate influence?
7. How well is the client represented within those same sources?
8. Which information is outdated, incomplete, incorrect, or unsupported?
9. What credible evidence does the client possess that is missing from the external information environment?

That is a much more useful diagnostic.

From Backlink Building to Citation Engineering

The phrase citation engineering needs careful definition.

I do not mean manufacturing citations.

I do not mean paying websites to repeat claims.

I do not mean building a disguised link network.

I mean deliberately engineering the conditions under which legitimate citations become possible.

That includes creating:

• original research,
• useful datasets,
• expert commentary,
• product documentation,
• public methodology,
• independent testing,
• comparative evidence,
• transparent pricing information,
• correction resources,
• credible case studies,
• and verifiable statistics.

Then getting that evidence in front of the sources that actually matter.

The distinction is important.

Spam asks:

“How can I place my brand here?”

Citation engineering asks:

“What evidence would make this source legitimately want to reference my brand?”

That is a far more durable strategy.

The Source Is More Important Than the Link

Traditional digital PR often evaluates a placement partly through the value of the backlink.

Machine Relations asks a different question:

What role does this source play inside the machine evidence environment?

That may matter even if the link itself is:

• nofollow,
• buried in a research page,
• inside a forum,
• part of an industry database,
• or otherwise unattractive by traditional link-building standards.

If the source repeatedly influences high-value AI answers, the commercial value may come from information influence, not link equity.

This is where CiteWorks' approach to LLM optimization differs from conventional backlink acquisition.

The target is not the hyperlink.

The target is the citation relationship.

Citation Rating Tells Us Which Sources Matter

At LLM Authority Index, we use the term Citation Rating to describe a proposed framework for identifying which sources occupy disproportionately influential positions inside AI citation and evidence networks.

Read the Citation Rating definition

Citation Rating does not merely count how often a source appears.

It considers variables such as:

• cross-model appearance,
• recommendation proximity,
• source independence,
• persistence,
• category specificity,
• and network position.

The core idea is simple:

A source cited frequently is not necessarily the source shaping the conclusion.

For Machine Relations, Citation Rating becomes a prioritization tool.

Instead of producing a list of 5,000 websites that mention a category, we want to find the much smaller set that appears to exert disproportionate influence over machine conclusions.

Brand Rating Tells Us Where the Client Stands

Finding the influential sources is only half the problem.

The next question is:

Which brands dominate those sources?

That is Brand Rating.

Read the Brand Rating definition

Brand Rating measures how strongly a company is positioned within the sources, claims, and evidence most influential to AI-generated conclusions.

This matters because mention volume can be deceptive.

Suppose:

Client

• 2,000 web mentions
• excellent traditional PR visibility
• strong backlink profile

Competitor

• 800 web mentions
• fewer overall placements

But the competitor appears positively across seven of the ten highest-Rating sources in the category.

The client appears in only two.

The client may have a bigger PR footprint.

The competitor may have the stronger machine evidence position.

Machine Relations tries to close that gap.

The Citation Rating Gap Changes Media Targeting

This creates what we call the Citation Rating Gap.

The gap exists when:

the sources a marketing team prioritizes

are materially different from:

the sources machines appear to prioritize.

For example, a PR team may have spent years pursuing:

• national business media,
• large lifestyle publishers,
• high-DR editorial sites,
• major news organizations.

But citation analysis reveals that AI recommendations in the category are repeatedly supported by:

• two specialist trade publications,
• one major Reddit community,
• a technical comparison site,
• an industry association,
• and one highly specific research report.

That should change the communications strategy.

Not because the large media outlets suddenly stopped mattering.

But because the team has discovered a previously invisible influence layer.

Machine Relations Is Not “Get Mentioned on Every AI-Cited Website”

That would be the lazy interpretation.

And it would recreate the exact problem that made traditional link building so spammy.

A brand should not look at a list of influential sources and immediately attempt to force itself into all of them.

Instead, it should ask:

Why is this source influential?

What does it know about our category?

What evidence does it use?

Where is its information incomplete?

Is our competitor genuinely better represented?

Are we absent because we haven't supplied useful evidence?

Is the source using outdated information?

Does our product actually deserve the claim we want associated with it?

Those questions force better marketing.

Sometimes Machine Relations will identify a communications opportunity.

Sometimes it will identify a data problem.

Sometimes it will identify a product problem.

Sometimes the AI may be right.

That is important.

If the Machines Are Right, Marketing Isn't the Solution

Suppose AI systems consistently say:

“Competitor A has better customer service.”

Citation analysis shows that:

• several independent review sites agree,
• customer discussions agree,
• third-party surveys agree,
• and the client's own ratings are weaker.

The correct Machine Relations recommendation is not:

“Let's manipulate the AI narrative.”

The recommendation is:

Improve customer service.

Then create credible evidence that the service improved.

AI optimization should not become a mechanism for hiding bad products.

The strongest long-term LLMO strategy is to make the underlying reality more citable.

Evidence Is the New PR Asset

This is one of the biggest changes I see coming.

Traditional PR prized:

• stories,
• quotes,
• announcements,
• executive access,
• exclusives.

Machine Relations adds another premium asset:

structured evidence.

Consider the difference between these two pitches.

Pitch A

“Our CEO believes our product has the best retention in the industry.”

Pitch B

“We analyzed 1.8 million anonymized customer events across 36 months. Here is the methodology, underlying aggregate dataset, confidence interval, and full result.”

The second pitch gives journalists something to cite.

It gives analysts something to investigate.

It gives publishers something to reference.

And potentially, it gives retrieval systems a clean evidence object.

That is citation engineering.

Original Research Has Compounding Value

Good original research can produce several layers of return.

It can:

1. Earn editorial coverage.
2. Generate conventional backlinks.
3. Create brand mentions.
4. Become a source for future articles.
5. Enter AI retrieval environments.
6. Influence category narratives.
7. Produce future human citations when people discover it through AI.
8. Strengthen the brand's association with the topic.

This connects directly with the Algorithmic Reciprocity Loop developed at LLM Authority Index.

Read: The Algorithmic Reciprocity Loop

The proposed loop is:

machine recognition

→ human discovery

→ human citation

→ conventional web authority

→ stronger future discoverability

→ additional machine recognition

Original research is particularly well positioned to participate in that cycle because it gives both humans and machines a reason to reference the original source.

This Is Why Indexes Are Powerful

One of the strongest content formats in Machine Relations may be the index.

Indexes convert a messy market into a measurable artifact.

Examples include:

• fastest-growing companies,
• safest cities,
• most trusted brands,
• market share indexes,
• compensation benchmarks,
• sentiment indexes,
• citation indexes,
• AI visibility indexes.

An index creates something that did not exist before:

a new measurement.

That makes it inherently citable if the methodology is credible.

At LLM Authority Index, we have been exploring this through the Consensus Index framework.

Read: The Consensus Index

The purpose is not to say:

“AI consensus equals truth.”

The purpose is:

to make machine consensus observable.

That creates data journalists, companies, analysts, and AI systems can potentially reference.

Aging in Place Index Is a Practical Example

One experimental application has been Aging in Place Index.

Instead of publishing a conventional “best medical alert systems” article based on one editorial opinion, the project compares how multiple AI systems evaluate the category.

It then preserves:

• model recommendations,
• ranking differences,
• supporting claims,
• disagreements,
• and cited sources.

That creates two useful maps.

The Recommendation Map

Which brands do the machines favor?

The Influence Map

Which sources appear behind those recommendations?

The second map is where Machine Relations becomes commercially important.

If certain sources repeatedly influence the category, brands need to understand those sources.

Not to manipulate them.

To understand the information environment that is already shaping machine conclusions.

Machine Relations Changes Competitive Intelligence

Traditional competitor research asks:

• Where do they rank?
• Who links to them?
• Where are they mentioned?
• What keywords do they target?
• What ads do they run?

Machine Relations adds:

• Which machine answers do they dominate?
• Which claims are associated with them?
• Which sources support those claims?
• Which high-Rating sources mention them?
• Where are we absent?
• Which sources mention both brands?
• Which claims differentiate them from us?
• Which pieces of evidence appear to drive their recommendation advantage?

That creates a very different competitive map.

The Goal Is Upstream Intelligence

Most reputation-monitoring systems work downstream.

They tell you:

“Here are 12,000 mentions of your company.”

That is useful.

But Machine Relations wants to move upstream.

If 800 of those mentions repeat the same claim, we ask:

Where did the claim originate?

If AI systems repeat the claim, we ask:

Which source did they retrieve?

If one influential publication sits behind dozens of downstream references, that source deserves attention.

Machine Relations is therefore partly a practice of information-source forensics.

It traces conclusions backward.

One Corrected Source Can Be Worth More Than 100 New Mentions

Imagine an important industry publication has outdated pricing for a company.

Other comparison sites repeated it.

AI systems retrieve the same information and tell users:

“Company X is significantly more expensive than competitors.”

The marketing team could respond by creating 100 new pages saying:

“We are affordable.”

That might accomplish very little.

A Machine Relations strategy would investigate the provenance.

Perhaps the outdated claim originates from one highly central comparison page.

Correcting that page with verifiable current information may be more strategically valuable than generating dozens of new low-influence mentions.

This is not link building.

It is evidence correction.

Machine Relations Is Also Reputation Management

The same method applies to negative narratives.

Suppose AI systems repeatedly mention:

“Brand X has difficult cancellation policies.”

There are several possible realities.

Scenario 1: The claim is accurate.

Fix the policy.

Scenario 2: The policy changed.

Update public documentation and communicate the change to relevant sources.

Scenario 3: The claim is incorrect.

Provide clear supporting evidence to publishers carrying the error.

Scenario 4: The claim is nuanced.

Create better documentation that explains the conditions accurately.

The objective is not to suppress criticism.

The objective is to increase information accuracy.

That is a much more defensible form of reputation management.

The Machine Relations Workflow

At CiteWorks Studios, I believe the emerging workflow looks something like this.

Step 1: Map Machine Opinion

Query the relevant AI systems across a controlled set of commercial and informational prompts.

Measure:

• recommendation frequency,
• comparative rankings,
• brand attributes,
• positive claims,
• negative claims,
• and uncertainty.

Step 2: Reconstruct Citation Provenance

Identify:

• cited URLs,
• domains,
• original sources,
• first-party versus third-party evidence,
• repeated claims,
• and source dependencies.

Step 3: Measure Citation Rating

Determine which sources appear most influential across the relevant query environment.

Step 4: Measure Brand Rating

Determine how strongly the client and competitors are represented inside those sources.

Step 5: Identify Evidence Gaps

Find:

• missing third-party validation,
• outdated information,
• unsupported claims,
• competitor narrative dominance,
• weak source representation,
• and opportunities for original research.

Step 6: Build Citation-Worthy Assets

Create material worth referencing:

• research,
• indexes,
• studies,
• expert commentary,
• technical resources,
• transparent datasets,
• comparative analysis,
• and useful tools.

Step 7: Conduct Targeted Human Outreach

Reach the journalists, publishers, communities, associations, creators, and experts who maintain the relevant information environments.

This part remains profoundly human.

Step 8: Re-Measure

Track:

• changes in citations,
• changes in Brand Rating,
• changes in model consensus,
• changes in recommendation share,
• and changes in the underlying evidence graph.

This turns LLMO from a collection of anecdotes into a measurable communications program.

Machine Relations Does Not Replace PR

This distinction is important.

I do not believe journalists disappear.

I do not believe earned media disappears.

I do not believe expertise disappears.

I do not believe relationships disappear.

Machine Relations actually makes good PR more valuable.

If machines rely heavily on third-party evidence, then credible independent publishers become critical parts of the information system.

What changes is our ability to measure which relationships have downstream machine influence.

Traditional PR asks:

Who influences people?

Machine Relations adds:

Who influences the information machines use when influencing people?

That is a second-order influence problem.

Machine Relations Does Not Replace SEO Either

Search still matters.

Crawlability matters.

Indexability matters.

Links matter.

Site architecture matters.

Google's own guidance for AI features says traditional Search fundamentals continue to apply to eligibility for its AI experiences.[2]

But eligibility is not the whole problem.

A company can have an excellent website and still be poorly represented in external evidence.

That is why Machine Relations is primarily an off-site discipline.

The company's own site provides the factual foundation.

The broader web determines whether other sources validate, challenge, or ignore those claims.

The Most Powerful LLMO Work May Happen Off-Site

This is an important strategic shift.

Much of SEO traditionally focuses on the client's website.

Machine Relations focuses heavily on the information environment surrounding the website.

That environment includes:

• publishers,
• forums,
• trade organizations,
• review sites,
• databases,
• research publications,
• social communities,
• expert commentary,
• videos,
• and independent testing.

The company does not control those surfaces.

That is precisely why they are valuable.

Independent evidence carries a different kind of credibility from owned copy.

Citation Engineering Is Not Link Engineering

The distinction deserves repetition.

Link engineering asks:

How do we create more links?

Citation engineering asks:

How do we create more legitimate reasons for influential sources to reference our evidence?

Sometimes the result is a link.

Sometimes it is an unlinked mention.

Sometimes it is inclusion in a table.

Sometimes a journalist quotes the research.

Sometimes a Reddit user references the finding.

Sometimes a reviewer updates a product comparison.

Sometimes an AI system later retrieves one of those pages.

The goal is broader than link equity.

It is evidence propagation.

Generative Engine Research Supports the Importance of Evidence Presentation

The academic paper GEO: Generative Engine Optimization examined how source visibility within generative-engine responses could change when publishers modified how information was presented.[3]

Among the techniques studied were the use of:

• citations,
• quotations,
• statistics,
• and other content characteristics.

The research does not provide a universal recipe for influencing every modern AI system.

But it establishes something important:

Source visibility within generative responses can be measured, and evidence presentation can affect that visibility.

For Machine Relations, that means publishers and brands should think not only about whether information exists, but whether it is:

• clear,
• attributable,
• specific,
• verifiable,
• and easy to understand.

Citation Presence Is Not the Same as Citation Quality

Research evaluating citations in generative search also gives us an important warning.

Nelson Liu and colleagues studied the verifiability of generative-search outputs, examining whether citations actually supported the claims attached to them.[4]

This matters because Machine Relations cannot operate on the assumption that:

“AI cited a source, therefore everything is correct.”

Citation provenance needs to be audited.

A source can be:

• relevant but misinterpreted,
• accurate but outdated,
• authoritative but quoted out of context,
• first-party but presented as independent,
• or simply wrong.

The objective is not to maximize citations blindly.

It is to improve the quality of the evidence network.

The Commercial Opportunity Is Measurement

The first phase of LLMO has largely been:

“Does ChatGPT mention us?”

The second phase will be:

“How often?”

The third phase will be:

“Why?”

That third question creates the real strategic value.

Once we can answer why, companies can allocate resources intelligently.

They can determine:

• which publications matter,
• which narratives matter,
• which evidence is missing,
• which sources need correction,
• which competitors dominate,
• and which campaigns actually change the machine evidence environment.

That is when LLM optimization becomes an executive-level discipline rather than another visibility dashboard.

The New PR KPI Is Not “Placements”

Imagine a campaign produces:

74 media placements.

That sounds impressive.

But now compare it with another campaign:

11 placements.

Those 11 placements produced:

• four appearances in high-Citation-Rating sources,
• two new independent product validations,
• three high-value comparative mentions,
• an increase in Brand Rating,
• and a measurable increase in AI recommendation share.

Which campaign was more valuable?

Placement count alone cannot answer.

Machine Relations requires a new KPI layer.

Possible measurements include:

• high-Rating placements,
• independent evidence growth,
• Brand Rating movement,
• AI recommendation share,
• query-level recommendation coverage,
• citation diversity,
• positive recommendation proximity,
• and persistence.

That is much closer to business impact.

The Future PR Agency May Look More Like an Intelligence Firm

This is the larger transition I expect.

PR agencies traditionally employ:

• media strategists,
• writers,
• publicists,
• account managers,
• relationship builders.

The Machine Relations agency adds:

• citation analysts,
• provenance researchers,
• data scientists,
• LLM visibility analysts,
• source-network researchers,
• and evidence strategists.

The output is no longer simply:

“Here are the journalists we should pitch.”

It becomes:

“Here is the evidence network shaping machine opinion in your category, here is where your competitors dominate it, here are the central sources, and here is the evidence required to legitimately improve your position.”

That is a substantially more strategic product.

From SEO to LLMO to Machine Relations

I see the evolution roughly like this:

SEO

Optimize the website to earn search visibility.

Digital PR

Earn authoritative third-party coverage and links.

LLMO

Improve visibility and representation inside generative systems.

Machine Relations

Understand and influence the external evidence environment from which those systems form conclusions.

Machine Relations is therefore not a replacement for LLMO.

It is one of the operating disciplines required to do LLMO properly.

The Research Layer and the Execution Layer

This is why we separate our work into two complementary organizations.

LLM Authority Index

Develops and publishes the measurement frameworks.

These include:

• The Consensus Index
• Search Authority vs. Machine Authority
• Citation Rating
• Brand Rating
• The Algorithmic Reciprocity Loop

CiteWorks Studios

Applies that intelligence commercially.

Our focus is not merely on securing mentions.

It is on understanding which citations matter, why they matter, and what legitimate evidence can earn them.

Research maps the network.

Machine Relations operates inside it.

The Question Every CMO Should Ask

For years, CMOs asked their SEO teams:

“Where do we rank?”

Then they asked:

“How many backlinks do we have?”

Now many are beginning to ask:

“What does ChatGPT say about us?”

That is progress.

But it is still only looking at the output.

The more important question is:

“Which sources and claims are causing AI systems to say that about us?”

Once you can answer that, AI visibility becomes much less mysterious.

You can map it.

You can measure it.

You can identify gaps.

You can improve the underlying evidence.

And then you can measure again.

That is Machine Relations.

Machine Relations Is Ultimately About Truth Distribution

There is a version of AI optimization that becomes another race to manipulate algorithms.

I am not interested in that version.

The durable opportunity is much more valuable.

Companies already possess enormous amounts of information that the public information ecosystem does not understand properly.

Sometimes:

• the pricing changed,
• the product improved,
• the testing is stronger than people realize,
• important research was never distributed,
• outdated claims still circulate,
• or competitors simply communicate their evidence more effectively.

Machine Relations is the discipline of making accurate, useful, verifiable information more available where it matters.

When the underlying reality is weak, fix reality.

When the evidence is weak, create evidence.

When the evidence exists but is invisible, distribute it.

When the evidence is wrong, correct it.

That is citation engineering at its best.

From “Who Will Link to Us?” to “Who Shapes the Answer?”

That is the transition.

Traditional link building asks:

Who will link to us?

Digital PR asks:

Who will talk about us?

Machine Relations asks:

Who shapes the answer?

Those questions are not mutually exclusive.

But the third one changes the strategy.

Because in an AI-mediated discovery environment, the most valuable source may not simply be the website with the largest audience or strongest SEO metric.

It may be the source sitting closest to the machine's conclusion.

Understanding those sources—and earning legitimate inclusion within their evidence—is what I believe the next generation of public relations will increasingly look like.

That is why PR is becoming citation engineering.

Sources and Research

1. Lewis, Patrick et al. — “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.” NeurIPS, 2020.
https://arxiv.org/abs/2005.11401

Foundational research demonstrating the combination of external document retrieval with language generation. Commercial AI-search systems differ significantly in implementation, but the work provides important technical context for understanding why external evidence sources matter.

2. Google Search Central — “AI Features and Your Website.”
https://developers.google.com/search/docs/appearance/ai-features

Google's official publisher guidance explains that established Search fundamentals remain relevant to appearing within Google's AI-powered search experiences. It reinforces that AI visibility remains connected to the broader web and conventional search infrastructure.

3. Aggarwal, Pranjal et al. — “GEO: Generative Engine Optimization.” arXiv:2311.09735; presented at KDD 2024.
https://arxiv.org/abs/2311.09735

One of the foundational academic studies treating source visibility within generative-engine responses as a measurable outcome. It provides evidence that content and evidence presentation can affect visibility in generative environments, although it should not be interpreted as a universal optimization formula.

4. Liu, Nelson F. et al. — “Evaluating Verifiability in Generative Search Engines.” 2023.
https://arxiv.org/abs/2304.09848

Research examining whether citations used by generative-search systems adequately support the claims attached to them. This provides an important evidence base for emphasizing citation provenance and evidence quality rather than citation quantity alone.

5. Gao, Tianyu et al. — “Enabling Large Language Models to Generate Text with Citations.” 2023.
https://arxiv.org/abs/2305.14627

Research on citation-supported language generation. Relevant to the broader emergence of citation and source attribution as measurable components of generative systems.

6. Page, Lawrence; Brin, Sergey; Motwani, Rajeev; Winograd, Terry — “The PageRank Citation Ranking: Bringing Order to the Web.” Stanford InfoLab, 1999.
https://ilpubs.stanford.edu:8090/422/

PageRank provides historical context for why relationships among information sources can matter more than simple counts. Machine Relations extends this general network perspective toward AI evidence environments rather than treating all placements equally.

7. Kleinberg, Jon M. — “Authoritative Sources in a Hyperlinked Environment.” Journal of the ACM, 1999.
https://doi.org/10.1145/324133.324140

Kleinberg's distinction between authorities and hubs provides useful conceptual precedent for thinking about the different roles sources can occupy within information networks.

Methodology and Disclosure

Machine Relations is a working strategic framework proposed by Mark Huntley and CiteWorks Studios.

It is not a documented discipline or ranking system used by Google, OpenAI, Anthropic, Perplexity, Microsoft, or another AI provider.

The framework combines established ideas from:

• public relations,
• digital PR,
• network science,
• information retrieval,
• citation analysis,
• reputation management,
• and generative-search research.

Machine Relations is built around a testable premise:

Brands can improve their AI representation most sustainably by improving the quality, accuracy, distribution, and independent validation of the evidence available within sources that influence machine-generated conclusions.

CiteWorks Studios does not treat AI citations as equivalent to conventional backlinks and does not assume that a citation within an AI interface transfers traditional PageRank.

The focus is on the broader information environment and the second-order effects that can occur when machines, publishers, companies, and users repeatedly interact with the same evidence.