As agentic AI begins to drive online discovery and decisions, Cimulate’s cofounders—an MIT professor and a retail tech veteran—are building the simulation-powered engine that could level the playing field for every brand not named Amazon.
The future of shopping won’t start with a keyword—it will start with a prompt.
As Agentic AI begins to drive how consumers discover and decide what to buy, Cimulate’s cofounders—an MIT professor and a retail tech veteran—are building the simulation-powered engine designed to level the playing field for every brand not named Amazon.
This week, the Boston-based startup launched CommerceGPT, an AI-native context engine trained on synthetic shopping behavior, not just historical clicks. Its goal: to help retailers show up, convert, and thrive in a world where agents—not people—drive discovery.
The shopping interface has changed. Cimulate wants to power what comes next.
For over a decade, digital commerce revolved around keyword search. But as Agentic AI systems like ChatGPT, Claude, and Perplexity begin driving how consumers discover, decide, and transact, a new model is emerging—one where shopping starts with a prompt, and the agents do the rest. Cimulate is building the infrastructure for that future.
That’s the world Cimulate is built for.
CommerceGPT doesn’t rely on brittle rules or clickstream data. Instead, it simulates millions of shopping journeys, creating synthetic behavior that trains its models to understand nuance, intent, and conversion at scale.
“Retailers shouldn’t need Amazon’s data to compete with Amazon,” says cofounder and CEO John Andrews. “CommerceGPT gives every brand an intelligence layer they’ve never had before.”
The Founders: From MIT’s AI Labs To Retail’s Next Operating System
Cimulate’s ambition is matched by its pedigree.
Cofounder and CTO Vivek Farias is a renowned MIT professor whose research in reinforcement learning and applied probability has significantly contributed to shaping the field of intelligent systems. His academic work was influential in the development of the company’s proprietary simulation method, which generates high-fidelity synthetic data tailored to each brand’s product catalog and shopper context.
Farias partnered with John Andrews, a retail tech operator with deep roots in ecommerce infrastructure. Andrews was an early executive at Endeca, the discovery platform acquired by Oracle for over $1 billion, where he later led commerce innovation across the cloud portfolio. Previously, Andrews and Farias cofounded Celect, a predictive analytics startup that brought AI to inventory optimization and was later acquired by Nike—an exit that underscored their ability to turn academic innovation into enterprise impact.
The two had long shared a frustration: retailers were being sold “intelligent” search and personalization, but the underlying tech was brittle, rules-based, and blind to context.
“We saw retailers being sold dashboards and rule engines dressed up as AI,” says Andrews. “We decided to build real intelligence—starting with the shopper.”
In 2023, they founded Cimulate, assembling a senior team with deep technical and commercial DNA.
“We realized the future of commerce wouldn’t be driven by better filters or keywords,” says Farias. “It would be driven by agents that can reason, recall, and recommend—and by the infrastructure that trains them.”
Together, the team is building Cimulate not as a feature layer, but as the foundation of a new commerce stack.
Why Agentic AI Is Reshaping How Consumers Shop
Cimulate is building the first AI-native platform for the rapidly evolving field of introducing a new term—and a new category of agentic commerce. The idea is simple but profound: when LLMs become the starting point for shopping, brands must optimize not for search engines, but for AI agents.
And that requires a different kind of infrastructure.
CommerceGPT is trained not just on past behavior, but on simulated behavior—millions of “what if” scenarios generated by large language models and refined through reinforcement learning. This enables the platform to deliver context-aware product recommendations, real-time search results, and AI-native discovery flows—even for retailers without extensive user data.
The system is already being used by brands such as PACSun, Tillys, and West Marine, with early results showing sharp increases in engagement, time-to-value, and conversion.
“Cimulate stood out because they didn’t just promise better search—they delivered a next-generation discovery experience tailored to our brand, our products, and our customers,” said Erik Quade, Chief Information Officer, Tillys. “Their AI-native approach gives us the ability to engage shoppers with precision, speed, and relevance in a way that legacy systems simply can’t match.”
How Cimulate Powers Prompt-to-Purchase With Agentic AI
Alongside its discovery engine, Cimulate is launching the MCP Server—a second layer of infrastructure designed for a world where agents communicate directly with each other.
As consumers increasingly issue open-ended prompts—“Find me a carry-on that fits JetBlue specs,” or “What’s a good reef-safe sunscreen?”—the winning product isn’t the one with the best metadata. It’s the one that shows up in conversational results.
MCP Server allows brands to create agents that speak the language of answer engines—optimizing not for Google keywords, but for LLM-native interactions inside ChatGPT, Claude, and Perplexity.
“We’re going from SEO to AEO—Answer Engine Optimization,” says Andrews. “Retailers need to stop writing for algorithms and start writing for agents.”
From SEO to AEO: Why Answer Engines Are the New Search Bar
AEO, or Answer Engine Optimization, is an emerging strategy for making content and products discoverable by AI agents—not search engines. As consumers increasingly rely on tools like ChatGPT to make decisions, brands must ensure their products are contextually surfaced within AI-generated answers, rather than being buried in search results.
The Competitive Landscape: How Cimulate Stacks Up In Agentic AI Commerce
Cimulate isn’t the only company betting on AI-driven commerce, but its approach is one of the boldest.
Companies like Constructor, Algolia, and Bloomreach offer AI-powered search and personalization. Meanwhile, synthetic data startups like MostlyAI and Tonic.ai help enterprises train models without risking privacy. And at the platform level, Amazon’s Rufus, Google’s Search Generative Experience, and Shopify Magic are embedding LLMs into native commerce flows.
Cimulate isn’t trying to out-Amazon Amazon. It’s building for everyone else—the brands that still need to be found, chosen, and bought inside the new discovery interfaces.
In that sense, Cimulate’s positioning echoes Daydream, the AI shopping startup founded by Julie Bornstein, which is also embracing agentic commerce. However, Daydream owns the frontend fashion experience, while Cimulate powers the backend intelligence stack across various verticals.
What Makes Cimulate the Infrastructure Layer for Agentic AI
- Simulation over rules: Instead of waiting for data, Cimulate generates high-fidelity synthetic shoppers and scenarios that train models on what drives conversion.
- Built for agents, not just humans: Most personalization engines optimize for clicks. Cimulate builds for LLM agents who shop based on context, not historical browsing.
- LLM-native from day one: With MCP Server, Cimulate helps brands show up in answer engines—not just search results.
This isn’t AI bolted onto ecommerce. It’s AI as infrastructure.
The company is backed by investors including Spark Capital, Sierra Ventures and Pillar VC, and has quietly built a team of top AI researchers, search engineers, and commerce leaders.
According to Spark Capital Co-founder and General Partner Alex Finklestein, Cimulate’s architecture solves a fundamental gap in modern commerce: the lack of infrastructure truly native to how LLMs operate.
“At Spark, we invest in both ends of the AI spectrum—from frontier model development with companies like Anthropic, to real-world, high-impact applications like Cimulate. What impressed us about Cimulate was their ability to turn cutting-edge AI into tangible business outcomes for retailers,” says Finkelstein. “They’ve built a platform that doesn’t just bolt on AI—it reimagines the entire product discovery experience for the agentic era. That’s the kind of category-defining company we look to back.”
Cimulate’s Long Game: Becoming the OS for Agentic AI Commerce
For retailers navigating the rise of Agentic AI, Cimulate is more than a feature layer. It’s the infrastructure layer for agentic commerce, which powers how discovery happens—across agents, interfaces, and intent.
Andrews frames it this way: “We’re building Stripe, but for discovery.”
Just as Stripe removed the friction from payments, Cimulate aims to remove the friction from AI-native (LLM-based) discovery. Its simulation-first engine turns synthetic behavior into search relevance, and prompts into purchases.
In an ecommerce future where the first move is made by a machine, Cimulate helps ensure brands are not just found—but chosen. It provides the foundation for retailers to be discovered, chosen, and purchased in a world where machines—not people—drive the first move.
