So yeh. The future of digital interactions is moving from "how" to "what" – from specifying steps to declaring outcomes.
ChatGPT Pulse proactively delivers personalized updates based on your chats, feedback, and connected apps, anticipating needs rather than waiting for commands. Similarly, Uniswap’s cross-chain intents allow users to specify the end-state of a crypto trade – and a network of fillers compete to fulfill the user’s outcome as fast and cheaply as possible.
These aren’t isolated innovations; they signal a broader shift toward intent-driven systems.
And here’s what makes this urgent: soon it won’t just be humans pushing these intents – it’ll be AI agents acting on their behalf. When millions of agents start transacting simultaneously, the scale and speed of commerce could explode exponentially. Platforms that aren’t ready for machine-speed markets may struggle to compete.
From Cart Management to Purchase Fulfillment
Today’s e-commerce requires customers to navigate multiple decisions: finding products, comparing options, checking inventory, selecting shipping methods, entering payment details. It’s a process designed around the merchant’s catalog structure, not the customer’s actual intent.
An intent-driven e-commerce platform would flip this model. Imagine customers expressing:
- "Find me a gift under $50 for my nephew who loves dinosaurs"
- "I’ll buy a Gozney Dome but only when it’s on 20% sale"
- "I really want the next Travis Scott Nike sneaker drop"
- "Tell me only when you have deep discounts on organic bedding, not your other sales"
- "Alert me for size 12 running shoes under $100, but only brands you know I like or want to try"
- "I need this today but with buy-now-pay-later options"
- "Show me merchants who offer wholesale rates for bulk orders"
The system would then handle the complexity – searching inventory, optimizing for price and delivery constraints, bundling items efficiently, even splitting orders across fulfillment centers if needed.
This is the paradigm shift: merchants would compete to fulfill customer intents based on their unique capabilities and specializations.
Supply-Driven vs. Demand-Driven Commerce
Think about what this really means. Today’s e-commerce is supply-driven: "I have these products, now I need to find customers." Merchants invest in SEO, ads, email campaigns – all pushing their inventory out into the world, hoping to intercept customer demand.
Intent-driven commerce flips this entirely. It becomes demand-driven: "Customers have these needs, now I need to win their orders."
Instead of asking "How do I get traffic to my store?", merchants ask "What intents can I fulfill better than anyone else?"
It’s a fundamental shift from marketing products to competing on capabilities – and it requires an entirely different skillset.
The Mental Models Behind This
I’m synthesizing several frameworks to understand this shift:
-
Jobs-to-be-Done (Clayton Christensen) – Customers don’t want products, they want outcomes. Intent-driven commerce finally aligns systems with what customers actually hire products to do.
-
Disruptive Innovation (Clayton Christensen) – This follows the classic disruption playbook: intent-driven systems start simple (basic queries) but will move upmarket to displace complex interfaces entirely.
-
Value Chain Analysis (Michael Porter) – But inverted: instead of optimizing internal operations to push products out, merchants optimize capabilities to pull customer intents in.
-
Platform Thinking (Parker, Van Alstyne, Choudary) – The platform creates network effects by connecting intent supply (customers) with fulfillment demand (merchants), with value increasing as both sides grow.
-
First Principles Thinking – Stripping away the accumulated complexity of shopping carts and checkout flows to the fundamental question: What does the customer want to happen?
The synthesis reveals something none address alone: when AI agents mediate commerce at machine speed, the entire competitive landscape inverts from supply-driven to demand-driven.
The Agentic Commerce Opportunity
This vision aligns with the broader movement toward agentic commerce – where AI agents act on behalf of users to complete complex tasks.
In this new world, success isn’t about driving traffic – it’s about fulfillment excellence and automation sophistication. Store owners could specialize in:
- Speed fulfillment: "I can deliver any intent requiring same-day delivery in urban areas"
- Budget optimization: "I specialize in fulfilling gift intents under specific price points"
- Subscription intelligence: "I excel at learning and anticipating recurring purchase patterns"
- Agent-to-agent readiness: "My systems can negotiate with AI agents, evaluate 10,000 intents per second, and commit in milliseconds"
- Payment flexibility: "I offer BNPL options on all orders over $50"
- B2B capabilities: "I can fulfill wholesale/bulk intents with net-30 terms"
Small merchants could compete on their unique strengths rather than trying to match Amazon on everything. The intent layer becomes an equalizer, letting specialized stores shine where they excel.
A local boutique might never rank on Google for "gift ideas," but they could dominate the "$50 nephew dinosaur gift" intent through superior curation and gift-wrapping – if they have the tools to compete at agent speed.
This represents a significant shift in e-commerce dynamics – from "build it and they will come" to "they’ve come, now compete to serve them best."
The Technical Foundation
To build an intent layer into e-commerce platforms, we need to think about:
1. Intent Recognition Engine
Integrate NLP capabilities to parse customer requests across multiple surfaces – whether they come through search, chat, email, or messaging integrations.
The system needs to understand natural language requests and translate them into structured intents that can be routed and fulfilled.
2. Merchant Capability Registry
Understand what each merchant can actually deliver – and this doesn’t necessarily mean asking them to fill out forms.
We could infer capabilities from actual performance data:
- Average shipping times
- Return rates
- Product range
- Peak capacity
- Available payment methods
- B2B pricing tiers
The catch? This requires deep, real-time connections to merchant stores.
The more data we have about actual merchant performance – not self-reported capabilities – the better we can match intents. Which payment gateways do they use? Do they offer BNPL? Can they handle net terms? Do they have volume discounts?
This creates a powerful incentive for merchants to fully integrate with platforms. The stores with the deepest connections get the best intent matches. It’s a virtuous cycle.
3. Fulfillment Orchestration
Platforms need to intelligently route intents to the best-matched merchants, potentially even splitting complex orders across multiple stores seamlessly.
But here’s the critical piece: merchants need agentic fulfillment commitment.
We can’t have humans manually accepting every routed intent. Merchants would set automated rules:
- "Auto-accept any gift order under $100 if I have inventory and capacity"
- "Accept same-day delivery requests only if order placed before 2pm and within my Zone A"
- "Bid on dinosaur toy requests but only if margin exceeds 30%"
- "Auto-accept wholesale requests over 50 units with net-30 terms"
- "Only accept BNPL orders if total exceeds my $75 minimum"
This means building merchant-side automation tools that let stores compete at machine speed.
If customer agents start submitting thousands of intents per second, only merchants with sophisticated automation will be able to participate effectively.
Think of it like high-frequency trading but for commerce – merchants set their parameters, and their systems automatically commit to fulfill intents that match their criteria.
The most sophisticated merchants might even use dynamic pricing, adjusting their bids based on current capacity, seasonality, or inventory levels.
4. Trust and Verification Layer
We need mechanisms to ensure intents are fulfilled as promised – automated quality checks, delivery confirmations, and smart dispute resolution.
And as much of this may be agent-to-agent, maybe we’ll see product-market fit for crypto solutions here. Smart contracts and on-chain verification could provide the trust layer needed for high-velocity, automated commerce.
While the underlying technology might vary – from REST APIs to GraphQL to event streams – the key is creating infrastructure where customer intent drives the entire transaction flow.
Privacy and Personalization Balance
Like ChatGPT Pulse synthesizes information from your memory, chat history, and direct feedback to learn what’s most relevant to you, e-commerce platforms have an opportunity to learn from purchase patterns while keeping privacy paramount.
Making personalization explicit and controllable:
- Standing preferences: Let customers set persistent rules: "Always show eco-friendly options first"
- Session contexts: Allow temporary intent modifiers: "I’m shopping for a party this weekend" that expire automatically
- Privacy controls: Granular toggles for what platforms remember: "Don’t track gift purchases" or "Forget this browsing session"
- Intent-based subscriptions: Replace spam with precision: "Only tell me about size 11 Travis Scott Nikes" instead of getting every sneaker newsletter
This isn’t about collecting more data – it’s about giving customers control over what they explicitly want platforms to remember to serve them better. And it’s about turning noisy marketing into valuable, intent-matched notifications.
Where Platforms Could Start Today
While building the full vision of agentic commerce will take time, platforms can start experimenting now:
- Intent-aware search: Enhance product search to understand context: "business dinner appropriate" not just keyword matching
- Conversational order modifications: Develop APIs that let customers change orders through natural language: "Add two more and ship to my office"
- Predictive reordering: Create tools for merchants to identify reorder patterns and surface "time to restock?" prompts at the right moment
- Intent subscription system: Build infrastructure for hyper-targeted conditional triggers. Stock alerts or newsletter subscriptions but far more granular: "Alert me when those Nike Dunks drop in size 10" or "Tell me only about 40%+ discounts on Japanese denim"
- Capability matching prototype: Start with a simple pilot matching "I need it today" intents with merchants who have same-day delivery in specific zones
- Automated fulfillment rules: Let merchants set basic auto-acceptance criteria: "Accept orders under $X with standard shipping"
Each experiment teaches platforms something about intent recognition, merchant capabilities, and customer expectations – building toward the larger vision incrementally.
The Agent-to-Agent Future
As AI agents become more sophisticated and customers grow comfortable delegating tasks to them, the expectation will shift from "show me products" to "solve my problem."
But here’s what’s really coming: it won’t just be humans expressing intents – it’ll be their AI agents doing it on their behalf.
Imagine thousands of AI agents simultaneously negotiating:
- "My user needs office supplies by Monday, optimize for sustainability and cost"
- "Source 500 units of product X, best price, delivered over 3 months"
- "Monitor all merchants for this specific vintage watch, auto-purchase if under $5,000"
The speed, complexity, and scale of commerce could go exponential.
Human-speed commerce might be replaced by machine-speed markets. If this happens, merchants won’t just need to handle individual intents – they’ll need systems capable of evaluating and responding to millions of agent-submitted intents per second.
The winners will be platforms that can operate at this scale while maintaining trust and quality.
Learning from DeFi: The Uniswap Intent Model
Uniswap’s cross-chain intent system provides a compelling blueprint for how this could work in commerce:
How Uniswap Intents Work:
- User declares desired outcome: "I want 1000 USDC on Polygon"
- Intent is broadcast to a network of "fillers" (competing services)
- Fillers compete to fulfill the intent optimally (fastest, cheapest route)
- Best execution wins, user gets outcome without managing complexity
Translation to Commerce:
- Customer declares intent: "I need running shoes, size 12, under $100, by Friday"
- Intent is broadcast to merchant network
- Merchants compete based on their capabilities
- Best match wins, customer gets outcome without shopping complexity
The key insight: competition happens at the fulfillment layer, not the discovery layer.
In DeFi, users don’t need to understand bridges, gas fees, or liquidity pools. In intent-driven commerce, customers shouldn’t need to understand inventory systems, shipping zones, or payment gateways.
ChatGPT Pulse: Proactive Intelligence
ChatGPT Pulse demonstrates another critical aspect: anticipatory action.
It doesn’t wait for you to ask – it proactively surfaces:
- Updates on topics you care about
- Changes in information you previously queried
- New developments in your areas of interest
Applied to Commerce:
- "That vintage watch you wanted just listed for $4,200 (under your $5k max)"
- "Your usual protein powder is 30% off today (your threshold)"
- "Nike announced the Travis Scott collab drops next Tuesday (you wanted to know)"
The platform learns what you care about and notifies you precisely when it matters – without spam, without noise, just signal.
The Managed Platform Imperative
Here’s a hard truth: self-hosted solutions will struggle with this future.
Why?
- Data aggregation requirements: Intent matching needs real-time access to inventory, pricing, capacity across thousands of merchants
- Performance demands: Agent-to-agent commerce operates at millisecond latency
- Continuous evolution: The intent recognition and routing algorithms need constant improvement
- Competitive dynamics: Merchants connected to the best intent networks win
For sophisticated stores with full tech teams, building custom solutions is possible. For the mass market, they’ll rely on managed platforms that provide:
- Real-time data aggregation
- Intent routing infrastructure
- Agent-speed APIs
- Trust and verification layers
This isn’t about control – it’s about capability. The platforms that can coordinate intent fulfillment across massive merchant networks at machine speed will define the next era of commerce.
The Path Forward
The shift to intent-driven, agentic commerce isn’t a distant dream – it’s already happening. Projects across the industry are tackling different pieces of this puzzle, from natural language interfaces to autonomous purchasing agents.
If this pattern continues, the key question becomes how quickly platforms can build the infrastructure to support it – though the timeline remains uncertain.
What’s needed now:
- Focused experimentation
- Learning from each iteration
- The courage to rethink fundamental assumptions about how commerce should work
Where I Might Be Wrong
This analysis assumes several things that could turn out to be incorrect:
I’m assuming that intent recognition will become reliable enough to trust at scale. If the technology for understanding what users actually want remains clumsy or error-prone for longer than I’m projecting, we might stay stuck with explicit interfaces. Privacy concerns could also limit how much context systems can use to understand intent.
I’m assuming that AI agents will reach sufficient reliability for autonomous purchasing. If consumer trust in agents develops more slowly than the technology, or if early failures create backlash, adoption could stall for years. The question "would you let an AI spend your money?" might have a different answer than "would you let an AI write your emails?"
I’m assuming that the economics work for merchants. Intent-driven systems might concentrate too much power in platform hands, or the cost of agent-speed infrastructure might exceed the value for most merchants. If only the largest merchants can afford to compete, this could reduce rather than increase merchant diversity.
An alternative view would emphasize that humans value the browsing experience itself – that discovery, comparison, and consideration are features, not bugs. Shopping might be less about efficiency than about the experience of choosing. Intent-driven systems might optimize away something consumers actually enjoy.
I could be wrong about the timeline. Christensen’s disruption playbook suggests disruption takes longer than technologists expect. Current e-commerce platforms might adapt rather than being displaced. The most likely outcome might be hybrid systems that combine intent-driven and traditional interfaces.
If I’m overestimating the pace of AI agent adoption, or underestimating the regulatory and trust barriers, the shift I’m describing could be a decade away rather than imminent. But if the pattern is real, understanding the direction helps inform today’s platform decisions.
Conclusion: Beyond Shopping Carts
The future of commerce isn’t about better shopping carts – it’s about eliminating the need for shopping altogether.
Customers declare what they want to achieve, their agents handle the rest, and the system makes it happen.
That’s the promise of intent-driven commerce. And the platforms that build for this future – where machine-speed markets replace human-speed commerce, where demand drives supply rather than vice versa, where specialization trumps generalization – will define the next era of digital commerce.
The pieces are already emerging:
- ChatGPT Pulse showing proactive, anticipatory AI
- Uniswap intents proving competitive fulfillment at scale
- AI agents gaining sophistication and trust
What remains is connecting these patterns, building the infrastructure, and giving merchants the tools to compete in machine-speed markets.
The shift from "how" to "what" appears to be gaining momentum. If it continues, the question becomes who builds the infrastructure to support it.
Published: September 29, 2024
Related Content:
- Web Summit 2024: Orchestration, Flow, and Dynamic Resource Allocation – Intent not instructions theme
- There Is No Host: Freedom Stacks – Operationalizing freedom through technology
- Intents All The Way Down – Foundation for intent-driven thinking
Referenced Technologies:
- ChatGPT Pulse – OpenAI
- Uniswap Cross-Chain Intents – Uniswap Labs
Source: Originally published September 2024 as strategic exploration of e-commerce futures, sanitized and enriched for public release December 2024.