10 predictions for 2025 for AI in commerce
AI is reinventing the infrastructure stack for shopping
I’ve been taking stock of major milestones for AI in commerce in 2024, and I’m increasingly convinced that the year’s story boils down to a single metric: 1000x.
That’s the factor by which cost of applying advanced AI models has plummeted since 2021. Back when GPT-3 was making waves, cost was around $60 per million tokens.
Now? We’re looking at $0.06, thanks to open-source and the crazy shipping velocity from foundational model makers!
So what used to be exclusively reserved for deep-pocketed giants with large research teams is now universally accessible. Intelligence & automation isn’t an exclusive resource anymore—it’s everywhere, and in many cases, it feels practically free.
And with this shift, something bigger has changed: the mood. The whole “e-commerce is slowing down” narrative is laughable. What we’re seeing now is a techno-optimism that I haven’t felt in years — we are in the “new renaissance”, fueled by a new set of algorithms & faster compute as our tools. Smaller operating teams better enabled by smarter automation.
AI is driving real value for commerce, and you can see it in the bottomline already :
Walmart had a 22% increase in U.S. e-commerce sales for Q3’24
Shopify's GMV for Q3’24 was $69.7 billion, a 24% increase from last year
Perhaps even more astounding is how quickly we’ve come to take AI for granted. Tasks that seemed like far-future projects for online brands —like real-time inventory optimization or hyper-personalized shopping journeys—are now becoming must-haves.
“Can we do it is now” changing to “When can we do it?”.. The bar for “innovative AI” keeps moving up!
Ecommerce tech is a prime launchpad for AI, platforms are elevating their AI offerings and specialized startups are appearing every week. And Bay Area remains the nerve center for AI frontier —from multi-modal recommenders to generative cataloging to agentic optimizers.
Over the past year, I’ve been in the court as a player - sometimes watching from the front row - as generative AI elevated product listings, virtual storefronts, and customer engagement.
But we’re only scratching the surface here.
I have my looking glass now set on 2025 - and here’s 10 developments that I believe will change the shape of commerce :
1. Hyper-Personalized Customer Experiences
80% of frequent shoppers only shop with brands that personalize the experience and 81% of consumers expect to see “just for me”. From Sephora, to Milani Cosmetics, to Charlotte Tilbury - personalization is everywhere.
What’s coming up?
Dynamic website interfaces that adjust in real time to individual customer preferences.
Predictive replenishment as we anticipate demand for household items.
Anticipate and address customer needs before they explicitly state them, creating a seamless and intuitive shopping journey.
2. Generative Content goes mainstream
Generative AI already took a huge step forward with how brands are now creating content, from product descriptions and marketing campaigns to catalog imagery and video ads.
In 2025, this will accelerate - the frontier is moving very quickly here:
The frontier → Custom made artificial influencers with a fully managed team that is dedicated to updating their looks and everyday voice. These will come with not just images, but reels, snaps, stories and chats
In 2024 we saw AI tools generating product imagery for virtual try-ons and augmented reality shopping experiences.
AI-written copy that aligns with brand voice, contextually adapting for different audiences and platforms.
AI-crafted videos and 3D models that allow customers to interact with products virtually, enhancing engagement and purchase confidence.
Hot Take : Generative AI agencies will be the new influencer agency.
3. Smart AI-Powered Promotions
Flat sales are so over. The days of blanket discounts and static promotions will be replaced by intelligent, AI-first promotional strategies.
These will leverage behaviourial segmentation, price sensitivity analysis, and competitor benchmarking to offer:
Personalized discounts that optimize for customer conversion without eroding margins.
Dynamic promotions that adapt in real time based on factors like inventory levels and user behavior.
Competitive pricing strategies that align with market trends while safeguarding profitability.
Dynamic pricing & promotions are no longer a far future - this is already starting to come to life - personalized coupons and offers have been the staple of marketplaces - it will start becoming a whole lot more common for brands in 2025.
4. Model Hubs and Specialized AI Ecosystems for Commerce
While foundational models like GPT or Llama have shown remarkable versatility, 2025 will see the rise of commerce-specific model hubs:
Curated “Commerce Brain” Repositories: These hubs will host specialized AI models pre-trained on retail data—from fashion to consumer electronics—ensuring new e-commerce entrants can access top-tier predictive and generative capabilities without building from scratch.
Open Collaboration: Model hubs will allow commerce experts to collaborate and publish insights and algorithms and for brands to purchase it off a marketplace.
Proprietary Advantage: Larger brands may opt to train private versions of these models with proprietary data, creating unique capabilities—like hyper-personalized recommendation algorithms or specialized fraud detection modules—that serve as enduring competitive moats.
Think huggingface, for ecommerce.
Commerce has its own datasets, it’s own evals and it’s own algorithms. A model marketplace and model ecosystem where open communities and experts can create and collaborate with scaled up retailers & brands is but inevitable.
5. Agents become your new intern
Autonomous AI agents will emerge as a transformative force in e-commerce.
These agents will:
Handle end-to-end tasks such as inventory management, customer engagement, and personalized marketing campaigns.
Collaborate with human teams to solve complex challenges and deliver improved outcomes.
Operate independently to analyze data, make decisions, and execute actions across the value chain.
Shopify and Salesforce are already leading the charge here by introducing agentic and AI-driven tools. Shopify’s intelligent sidekick, Salesforce’s Agentforce seem to be last year’s big AI moves in the space - this is gonna kick up a notch. Agents are gonna be your new teammate for everything commerce ops - from sourcing, to fulfillment, to marketing.
6. Smart Storefronts
Smart storefronts will revolutionize how brands engage customers by dynamically adapting to interactions in real-time.
These storefronts will:
Leverage code generation models integrated with website and app builders to accelerate rendering and updates.
Offer personalized layouts and product displays tailored to individual browsing behaviors.
Integrate real-time analytics engine to continuously optimize the user journey
We are already seeing Shopify make progress in this domain with app blocks - apps that connect to storefront blocks and are rendered on the fly.
But this is just a start - the whole storefront will become generative and smart and we will have a connected insights engine powering these apps.
From remembering your choices to proactively suggesting you the right selection to bringing the right offers that helps you decide better - storefronts are all set for a smart evolution.
7. In-Game Commerce
In-game purchases reached $70 billion+ in 2024.
Games are fast emerging as a new sales channel for e-commerce as brands embrace immersive experiences powered by generative media. In 2025, we will see:
Exclusive in-game merchandise and promotions designed to resonate with gamer communities.
Seamless integration of shopping experiences into virtual environments, enabling players to purchase items without leaving the game.
Generative quests and levels that enhances brand loyalty through engaging narratives tied to merch and the game storylines. Think a Fenty cave where you get an avatar redo.
This fusion of gaming and commerce will continue to push the boundary of what shopping truly is - a fun activity.
8. Robotic Warehouses and Touchless Logistics
From Robo Arms and Picking in warehouses, to inventory forecasting models, to realtime pricing strategies - autonomous operations will allow brands to:
Maintain optimal inventory levels by predicting demand fluctuations.
Dynamically adjust prices based on market conditions, competitor strategies, and consumer behavior.
Automate the boring - say goodbye to manual returns management, fraud detection, and customer support through conversational AI and machine learning algorithms.
Say goodbye to operational boring workflows. AI is automating away boredom in 2025.
10. From CDP → “Intelligent” Stacks
Data platform vendors in 2025 will position themselves as the “system of record” for commerce, akin to how CRMs became the system of record for customer interactions.
The difference? These platforms will go beyond transaction logs to also manage:
Multi-modal data: Real-time signals about user behavior — clicks, dwell times, wishlist additions—that inform predictive and generative AI models.
API first Ecosystems: Awareness of API first ecosystems will rise and brands will prioritize API first vendors, allowing specialized intelligence providers to plug directly into larger retail data hubs without extensive custom development. Platforms lacking API’s and integrations are on the way out.
Predictive and Prescriptive Analytics: Instead of static reports, executives will rely on AI “assistants” that recommend specific actions. Assistants for business intelligence from sourcing to shipping to marketing - we will see the rise of assistants for intelligence.
Large AI models are foundationally built around centralized data platforms - we will see the “system of intelligence” wrap the customer data platforms.
I call this the rise of intelligent stacks - where the system of record changes to a system of intelligence. Think “Databricks” for Commerce.
10. Commerce SaaS is roadkill for Agents
With complexity rising in commerce the software stack in commerce has kept multiplying - ads, analytics, CDP, direct marketing, website, content, personalization, payments, reviews, video, chat, offers, loyalty, inventory, cataloguing - the list is endless.
The result - commerce stack is fragmented, lots of operational complexity and the observability and lack of integrations has led to broken windows consistently.
Layer on the complexity of pricing models, revenue attribution and all of a sudden the technology stack feels overwhelming.
Hot take: Large commerce marketing saas will implode and 25% of all commerce saas will die.
AI-native startups will redefine commerce by building software stacks that are deeply integrated with foundational AI capabilities.
These startups will focus on:
Transitioning from traditional tooling to systems of intelligence that combine models, algorithms, and agentic applications.
Making multi-channel commerce more cost-effective and streamlined by selling outcomes rather than just tools.
Pro modes - Introducing intelligent assistants as a service, much like OpenAI’s move to offer ChatGPT Pro editions.
Your margin is my opportunity. AI-native startups will capitalize on this market inefficiency by offering platforms that focus on outcomes, positioning themselves as the next generation of commerce infrastructure.
Commerce is embracing the smarts
The integration of AI into e-commerce is no longer an incremental improvement; it represents a foundational shift in how brands run their businesses and connect with shoppers.
It’s a call for smarter commerce — where algorithms work hand in glove with human creativity to accelerate commerce.
2025 - here we come!
