As consumers increasingly shop for goods using artificial intelligence (AI) chatbots — and as AI agents are being developed to autonomously pay at checkout — traditional brand websites must adapt or risk being invisible in this new internet.
New Generation, or simply New Gen, is a San Francisco-based startup that aims to solve a core part of that problem.
Founded a year ago by Harvard Business School alumni Jonathan Arena and Adam Behrens, the startup is building AI storefronts – versions of brands’ websites that dynamically interact with AI chatbots and agents.
“Every technology wave has required a new commerce layer to be developed,” Behrens said in an interview with PYMNTS. “With the early internet, it was PayPal. In the 2010s it was Stripe. … Given the characteristics of AI and how it’s this new channel and new technology, there actually is a need for a new payment and commerce infrastructure.”
So they both left their jobs — Behrens was at Stripe and Arena was at Patreon – to start New Gen, which combines Arena’s design expertise with Behrens’ technical skills.
The result is the AI storefront, a dynamic interface that can usher in a “truly bespoke internet for every person, which has been the biggest vision for a lot of companies over the past decades,” Arena said. “Now, we feel like it’s finally going to be possible.”
For example, if a customer is researching yoga gear within ChatGPT, instead of getting content from static blogs on a retailer’s website, the chatbot can generate personalized offers in real time, provide precise product details, pay for them and perform other tasks.
Another possible use case is that the AI storefront can reflect fast-moving marketplace changes, such as adjusting product listings in light of tariffs or to take advantage of real-time buzz around a celebrity or trends.
New Gen has partnered with Visa as part of the payments giant’s Intelligent Commerce initiative to enable AI agents to make purchases on behalf of consumers. Arena said they are also doing pilots with “large, global brands” in the retail, fashion, travel, dining and other industries.
To be sure, a truly autonomous AI agent is still some ways off, according to a forthcoming PYMNTS Intelligence report. Most business deployments of generative AI today still need employees to ensure processes are accurate and stay on track. Even tasks ripe for pure automation, such as managing cybersecurity systems, still rely heavily on human involvement.
Read more: Exclusive: Agentic AI Vision Meets Reality in New PYMNTS Report
3-Way Conversation
Still, the foundation for agentic commerce needs to be set in place early, especially since the technology is advancing rapidly.
Visa, Mastercard and PayPal already are developing agentic commerce infrastructure that lets AI agents complete the checkout process for consumers. Meanwhile, other parts of the digital economy need to change as well for this entire ecosystem to work — including digital storefronts.
“There’s really a three-way conversation that has to happen right now: the payment networks, the brands, and then what we’re building, which is the infrastructure to unlock it,” Behrens said.
“The world that we think is going to happen is that you’ll have an agent on both sides of the transaction,” Behrens added. But traditional websites break when AI agents try to navigate them, struggling with pop-ups, poor search functionality, and checkout flows designed to prevent bot traffic.
The founders sought to solve two critical problems plaguing retail websites to get brands ready for AI-driven commerce.
First, many still rely on outdated product management systems that require manual tagging. “If somebody goes to your website and searches for Taylor Swift lipstick, somebody needs to manually add Taylor Swift as an attribute tag” on the relevant products, Arena said. “That’s a really interesting thing that AI can now automate at scale.”
The second issue is moving from keyword searches to conversation-based searches driven by AI chatbots. Behrens said 800 million people a week use ChatGPT, but retail websites still require shoppers to use keyword-based searches to find products. New Gen’s AI storefront aims to change a brand’s traditional search capabilities.
Read more: Visa, Mastercard, PayPal Fuel Agentic AI Commerce Boom
AI Subdomain for Every Website
To build the AI storefront, New Gen creates an “AI subdomain” — a dedicated microsite such as “ai.brand.com” designed to receive AI-driven traffic from tools like ChatGPT or website-browsing agents like Operator.
“We are completely avoiding a rip and replace,” Arena said. “It’s really tough for brands [to dismantle their traditional websites]. So we give them a completely separate experience that works out of the box.
“The experience a customer sees is generated on the fly in response to whatever they ask,” Arena added.
Their system reindexes a brand’s entire product catalog, combines structured and unstructured data such as product descriptions and social media posts, and uses AI models to create dynamic product pages and recommendations. This data pipeline is available through an API for other companies to access.
For image generation, New Gen uses Gemini 2.5. For prose and copy, the team relies on Anthropic’s Claude 4. For code and front-end generation, OpenAI’s models are preferred. “Our offline processes use a lot of the higher-tier OpenAI models but the in-line stuff, we use a lot of the Flash mini models” for quick responses, Behrens said.
Brands can control the experience through a merchant dashboard, setting preferences for tone of voice, which products to highlight, and seasonal merchandising strategies. Over time, the AI storefronts would integrate with marketplaces such as Shopify or WooCommerce.
Read more: Agentic AI Is Quietly Reshaping the eCommerce Funnel
Read more: Startup New Generation Builds Agent-Friendly Storefronts for Brands
Read more: Forget Browsing. The Agentic AI Internet of Tomorrow Will Find You First
