Swipe to navigate
Strategic Research Brief - September 2025

The 7-10 Year Evolution of the Web

From Scroll to Stream, From Search to Answers

The architecture of the internet is shifting beneath us. The question isn’t whether—it’s how fast.

How AI agents will reshape traffic, revenue, and trust across the internet

The Shift

We’re Leaving the Age of the Page

Three decades of the same loop:
1

Search — type keywords into a box

2

Scroll — scan a ranked list of links

3

Click — visit a page, maybe find what you need

That loop built the entire digital economy. Advertising, publishing, e-commerce, SEO—every business model on the web traces back to this three-step pattern. Now it’s breaking.

The Paradigm

From Scroll to Ask

The fundamental interface is inverting.

The Old Paradigm

  • User formulates a query

  • Engine returns ranked links

  • User evaluates which to trust

  • Discernment is distributed

The New Paradigm

  • User asks a question

  • Agent synthesizes an answer

  • Links serve as validation, not destinations

  • Discernment is centralized

Already happening: Microsoft Copilot in Edge, Perplexity AI, Arc Search, ChatGPT with browsing. The URL bar persists—but it’s no longer the dominant point of entry.

The Economics

The Ad Model Fractures

Three pillars of the digital economy, all under pressure.

SectorCurrent ModelAgent-Era ModelImpact
AdvertisingImpressions, clicks, pageviewsSponsored answers in agent responses25-45% CTR decline when AI summaries appear
ContentPages as destinationsStructured data feeds, API licensingUnlicensed content risks invisibility
RetailMulti-step funnelsSingle-turn conversational purchaseStorefronts become infrastructure

The redistribution: Revenue moves away from open web inventory toward platform-controlled agent surfaces. The middleman changes, but the dynamic doesn’t.

Commerce

E-Commerce Funnels Collapse

Traditional Funnel

Awareness → Interest → Consideration → Intent → Purchase

  • Multiple touchpoints across sessions

  • Brand differentiation at every stage

  • Storefront as competitive advantage

Agent-Mediated Commerce

Ask → Compare → Buy (single conversation)

  • Discovery, evaluation, purchase in one flow

  • Agent selects on specs, reviews, price

  • Storefront becomes fulfillment layer

The threat to brands: When an agent comparison-shops, the differentiator isn’t your homepage design or checkout UX. It’s structured data, reviews, and API accessibility. Retail becomes logistics.

The Analogy

Streaming Is the Right Analogy

Every content industry follows the same arc.

IndustryPre-DisruptionResistance PhasePost-Disruption
FilmBox office + DVD salesSued Napster, blocked NetflixLicensed to streamers, lost leverage
MusicAlbum sales + radioDRM, lawsuits, walled gardensPer-stream royalties, platform dependence
Web ContentPageviews + ad inventoryBlocking scrapers, suing AI companiesLicensing deals, structured feeds, API access

The lesson: Aggregators consolidate distribution power. Producers lose direct audience access. Revenue compresses. The web is having its “streaming moment”—AI agents are the next aggregators.

New Markets

The Secondary Economy

When the primary model fractures, secondary markets emerge.

Content Licensing

Publishers license archives to AI platforms

  • • OpenAI/News Corp deals
  • • Per-query revenue sharing
  • • Structured data as asset class

Agent Placement

Brands pay for position in agent responses

  • • Sponsored answers
  • • Preferred vendor APIs
  • • Agent-optimized product data

Verification Services

Trust infrastructure for agent-mediated content

  • • Source attribution layers
  • • Fact-checking APIs
  • • Provenance certification

The pattern: Every disruption creates losers in the primary economy and winners in the secondary. The question is which side of the transition you’re building for.

Epistemics

The Trust Problem

The most significant risk isn’t economic—it’s epistemic.

When scrolling disappears, so does comparison. The first answer an agent delivers becomes, by default, the accepted answer.

Authority once distributed across billions of pages is concentrated within a handful of platforms controlling ingestion and ranking.

The concentration of truth: We’re moving from distributed human discernment (imperfect but pluralistic) to centralized algorithmic synthesis (efficient but monopolistic). The tradeoff is real and underexamined.

Discovery

AEO Replaces SEO

Answer Engine Optimization is the successor to Search Engine Optimization.

DimensionSEO (Search Era)AEO (Agent Era)
GoalRank on page oneBe cited in the answer
CurrencyKeywords, backlinksStructured data, schema markup
OutputBlue linksSynthesized answers with citations
Competition10 slots per page1-3 cited sources per answer
ManipulationKeyword stuffing, link farmsData poisoning, schema manipulation

The compression: SEO had 10 results per page. AEO has 1-3 citations per answer. The winner-take-all dynamic intensifies by an order of magnitude.

The Risks

Gaming the Answer

New incentives create new attack surfaces.

Every optimization regime creates a manipulation regime. SEO had link farms. AEO will have its own dark arts.

1

Data Poisoning

Injecting misleading structured data to manipulate agent responses

2

Paid Inclusion

Undisclosed commercial relationships influencing agent citations

3

Authority Spoofing

Manufacturing signals that make sources appear more authoritative to ingestion models

4

Feedback Loop Capture

Optimizing content to be cited, which increases authority, which increases citation

The uncomfortable parallel: The same people who built SEO empires are already pivoting to AEO. The manipulation won’t be less sophisticated—it’ll be harder to detect.

Strategy

Strategic Implications: Platforms

Search Engines

Evolve or be disintermediated

  • • Integrate agent layers (Google AI Overviews)
  • • Defend ad revenue with sponsored answers
  • • Acquire or build agent-native products

Social Platforms

The feed survives—maybe thrives

  • • Identity + social graph = moat
  • • Algorithmic feed is already agent-like
  • • User-generated content resists synthesis

Cloud & Infrastructure

Picks and shovels play

  • • Compute demand explodes
  • • Agent hosting becomes a service category
  • • API gateway economics change

AI-Native Startups

Build the new stack

  • • Perplexity, You.com, character-based agents
  • • Vertical-specific answer engines
  • • Compete on ingestion quality + trust
Strategy

Strategic Implications: Publishers & Retailers

Publishers

  • License content to AI platforms

  • Build structured data feeds

  • Develop direct audience relationships

  • Invest in brand authority signals

Retailers

  • Optimize product data for agent consumption

  • Build agent-accessible APIs

  • Compete on fulfillment, not discovery

  • Develop proprietary data advantages

Marketplaces

  • Amazon already agent-ready (structured catalog)

  • Marketplace model survives as fulfillment

  • Discovery layer vulnerable to external agents

  • First-party data becomes the real moat

Strategy

Strategic Implications: Advertisers & Developers

Advertisers

The media plan rewrites itself

  • Budget shifts from display to agent-surface sponsorship

  • Attribution models break and must be rebuilt

  • Brand building becomes the hedge against agent commodification

  • Contextual placement in agent conversations replaces behavioral targeting

Developers

The stack pivots from front-end to data-first

  • Schema creation and structured data become core skills

  • API design matters more than UI design

  • Agent integration becomes a standard competency

  • Traditional front-end web development deprioritized

The skill shift: The web developer of 2030 spends more time on data architecture and agent integration than on CSS and JavaScript frameworks. The interface layer thins as the data layer thickens.

Counterpoint

Will Agents Replace the Browser?

The strongest version of the argument—and why it might be wrong.

The Bull Case for Replacement

  • Conversational interfaces are more efficient for information retrieval

  • Agents can act, not just answer—book, buy, schedule

  • Mobile already reduced browsing to app-hopping

  • Younger users default to AI for research

The Bear Case Against

  • Visual browsing (shopping, design, entertainment) resists text synthesis

  • Regulatory friction on agent-mediated transactions

  • Browser vendors will absorb agent features, not cede ground

  • Serendipity and exploration have value agents can’t replicate

My read: The browser doesn’t die. It evolves. Agent capabilities get embedded inside the browser, not outside it. Chrome with Gemini built in is more likely than a Gemini app that replaces Chrome.

The Exception

The Social Media Exception

Social media is the one part of the web that may be largely immune to agent disruption.

Why the feed survives:

  • • Identity is the product, not information
  • • Social graphs can’t be synthesized
  • • The feed is already an agent (algorithmic curation)
  • • User-generated content is the moat

Where agents do penetrate:

  • • Content creation (AI-generated posts)
  • • Comment moderation
  • • Ad targeting and placement
  • • Trend detection and curation

The strongest call in this deck: Social platforms survive because they sell identity and connection, not information. You can’t synthesize a friend’s opinion into an agent response. The feed is the one interface that agents augment rather than replace.

The Arc

The Web Unbundles and Rebundles

This isn’t new. The web has always oscillated between bundling and unbundling.

1

The Portal Era (1995-2005)

Yahoo, AOL bundled everything. One destination for news, email, search, shopping.

2

The Unbundling (2005-2015)

Google unbundled search. Craigslist unbundled classifieds. Vertical SaaS unbundled everything else.

3

The Platform Era (2015-2025)

Apple, Google, Amazon rebundled around ecosystems. Walled gardens with APIs.

4

The Agent Era (2025-?)

Agents rebundle everything into conversational interfaces. One entry point, infinite backends. The ultimate aggregation.

Timeline

The Strategic Calendar

Where we are and what’s coming.

PhaseTimelineWhat HappensKey Signal
Coexistence2024-2026AI summaries alongside traditional resultsGoogle AI Overviews, Bing Copilot
Tipping Point2026-2028Agent-first becomes default for information queriesPublisher licensing deals go mainstream
Restructuring2028-2031Business models fully reconfigure around agent distributionMajor publishers close or pivot to licensing
Maturity2031-2035New equilibrium: agents as primary interface, web as data layerAgent-native businesses outnumber web-native ones

Where we are now: Deep in the coexistence phase. The window for strategic positioning is open but narrowing. By the tipping point, the power dynamics will already be set.

The Truth Machine Crisis

“The early promise of the semantic web envisioned a distributed system of meaning. What is emerging instead is a commercially-driven, agent-controlled ecosystem.”

The next decade of the web will not be defined by the quality of answers—but by the consolidation of power among those who control the machines that deliver them.

Whether that’s dystopian or just the next iteration of how power always works on the internet—probably both.

Signal Dispatch Research | September 2025