Advertising, Global Expansion & AI
The underlying logic of the next growth system
- Entry: search / recommendation → AI answers / agents
- Supply: written and made by people → AI-generated, human-directed
- Execution: people click pages → AI completes the task
- Goal: higher ROI × greater scale
The three parties in advertising: the part that doesn't change
Advertising comes down to buying low and selling high: acquire traffic at lower cost, convert and monetize at higher efficiency.
Three roles have always existed in an advertising system.
Pays for traffic and results.
Owns distribution and user attention, and absorbs advertiser budget.
Takes part through attention, interest, behaviour, and spend.
The relationships between them have not disappeared: advertisers pay platforms, platforms supply traffic, advertisers reach users through content and experience, and users' interest, behaviour, and spend feed back into what both platforms and advertisers decide next.
AI did not kill advertising. It is moving advertising from "matching attention" toward "understanding intent and completing the task."
How the entry point evolved
The ad system matches around keywords, results pages, and explicit demand.
Platforms predict interest algorithmically and distribute content and ads in the feed.
AI handles longer sentences and vaguer needs, and answers or recommends directly.
An agent doesn't just answer — it understands the need, calls tools, and gets it done.
A new set of entry points and ad placements follows from this.
Search, recommendation, AI answers, and agents will all continue to exist. What migrates is how users express demand and how they get results.
The funnel shifts from clicking to executing
- Impression
- Click
- Landing page
- Sign-up / Download
- Purchase / Payment
Traditional advertising delivers a user to a page. Everything after that is mostly the user's own work.
- Understand need
- Search & compare
- Call tools
- Order / Book
- Done
An agent can string together actions that were previously scattered across multiple pages, products, and steps. The endpoint of the advertising and growth funnel will gradually shift from "deliver a person to a page" to "get the thing done."
Monetization models
The product is free; revenue comes from splash, feed, and rewarded video formats.
Revenue from virtual goods, features, memberships, items, or premium tiers.
From Plus to Pro — renewals rest on continuous value, sunk cost, and habit.
Built around commerce, travel, finance, and education: download, sign-up, purchase, repeat.
Once agents start completing tasks for users directly, more value has the chance to migrate from impressions and clicks toward transactions, outcomes, and ongoing service.
Three layers where AI affects advertising
AI reads longer sentences and more complex needs, improving accuracy in search, recommendation, and ad matching.
AI produces text-to-image, image-to-video, digital humans, and voiceover quickly, cutting content production cost sharply.
Agents monitor, diagnose, optimize, and report automatically, lifting media efficiency through human-AI collaboration.
AI's value isn't only in "how much more content it generated." It's in whether understanding, decision, execution, and feedback can be compressed into one system.
Brand Agent × Service Agent
One judgment matters most here: a brand must build its own in-house agent capability.
- Where does the brand sit against category benchmarks?
- Is recent creative fatiguing?
- Are current bids competitive?
- Which data and changes need attention right now?
- Which strategies fit the brand's long-term goals?
- Which categories are breaking out right now?
- Which creative and which creators are scaling?
- What changed in category benchmarks and platform rules?
- What's the next action worth executing?
A brand agent has to carry the company's internal business goals, operating data, brand principles, budget boundaries, and organizational context. A service agent's core value comes from cross-client, cross-industry know-how, category benchmarks, and continuous tracking of media policy, platform change, and external supply.
In the past these judgments depended heavily on communication between people and between organizations. High cost, high latency, and serious information loss along the way. Agents let data, experience, strategy, and the ability to act settle into digital assets that keep running.
Brand agent and service agent collaborate directly over A2A: one holds the full context of the client and the accounts, the other supplies outside industry insight and specialist capability, and together they carry out analysis, decision, execution, and feedback.
The brand agent should not be outsourced — a company has to own its own data, goals, and context. The service agent won't disappear either, because cross-client industry knowledge, benchmarks, and media supply still hold independent value.
Agentic growth has already started
The talk cited public case data from Google AI Max: advertisers who turn on the relevant capabilities typically get around 14% more conversions or conversion value at comparable CPA or ROAS; for campaigns that previously ran mainly on exact and phrase match, the lift can reach about 27%. These are Google's own figures — results will differ by advertiser.
That means agentic growth is already showing up inside ad platforms. AI is taking part in understanding, matching, creative, delivery, and optimization, and making the growth loop turn faster.
As generation capability spreads, content and creative will become increasingly commoditized. Brand, creative judgment, data, distribution, and execution quality will still create difference.
MarTech and AdTech converge
- User insight
- Content / creative
- CRM / lifecycle
- Marketing automation
- Media buying
- Bidding / attribution
- Creative / testing
- ROI optimization
An AI agent needs to understand users, brand, content, and business goals, and at the same time connect to media, accounts, bids, creative, and delivery results. Analysis and actions that used to sit apart in MarTech and AdTech get strung by the agent into one shorter growth loop.
How growth value pools migrate
| Layer | Value pool | Direction |
|---|---|---|
| L0 | Cross-border GMV | Still growing, but decelerating |
| L1 | Media budget | Still growing; platforms still absorb the main value |
| L2 | Services / creative dollar pool | Automation compresses price; under pressure or shrinking |
| L3 | AI software / tokens | New demand plus workflow migration; steepest slope |
| L4 | Selling outcomes | Still early, but closest to what clients genuinely want to pay for |
Production capability commoditizes fast. What stays genuinely scarce keeps migrating toward distribution, data, industry context, and accountability for outcomes.
Three tracks for Chinese companies going global
Core growth path: global user growth, paid, and advertising.
Core growth path: mature media buying, retention, and strong monetization.
Core growth path: lower marginal cost and fast globalization.
Chinese companies' strengths in product iteration, content production, media buying, and monetization have a chance to keep spilling over into global markets alongside AI.
AI made Build cheap; Traction is still scarce
AI is rapidly lowering the cost of product development and content production, and more people can Build faster. But shipping a product is not the same as getting growth. Brand memory, channel breakthrough, distribution, and repeat purchase are all still scarce.
The end goal: the growth flywheel
A growth system can be read as one continuous loop.
- Know the user better: keep reading user intent
- Match faster: connect the right content, product, and service
- Deploy faster: finish testing, optimization, and scaling sooner
- Cost less: bring CAC down
- Worth more: lift LTV and ARPU
Together these point at higher ROI, which in turn brings more budget, greater scale, and more data — and keeps the flywheel accelerating.
Closing
AI is changing the entry, the supply, the execution, and the measurement of advertising and growth systems all at once.
The three-party relationship between advertisers, media, and users still stands. But the interface is migrating from pages to answers, from clicks to outcomes, and from point tools to agents that keep running.
The capability that really matters next is connecting user understanding, business goals, industry knowledge, media supply, and execution into a growth system that can keep learning.
The interface is moving from pages to outcomes.
Cited in this piece
