Founded in 2023 by a team of e-commerce veterans, ShopVision gives retailers a live, AI-powered view of their competitive landscape. It tracks the top 150,000 e-commerce sites in North America daily, capturing everything from paid ads to email campaigns to real-time pricing changes. Getting there required rethinking how browser agents could work at that kind of scale.

Key takeaways

  • Data pipeline time cut from 13 hours to 3 hours
  • 1.3 million browser sessions run in the last 30 days
  • 1.2 billion data points collected since inception
  • $1M+ in ARR closed in the first nine months
  • Super Agent: ShopVision's AI browser agent that lets retail teams ask any business question in plain English and get answers backed by real-time competitive intelligence

The challenge: collecting visual data at scale, without the overhead

From day one, the ShopVision team knew browser agents would be central to everything they build. E-commerce moves fast. Pricing changes overnight, promotions run for 48 hours, and a competitor's email campaign can shift market dynamics in a weekend. Stale data is unhelpful and a liability.

ShopVision's core data pipeline started out as a sequential process: for each new competitor site, their system would crawl the catalog, classify products, generate a collection script, and validate the output before ingesting data. End to end, that process took 13 hours per site. Long enough to miss a flash sale or a pricing move that lasted a single morning.

While most data collection tools pull raw HTML, ShopVision relies on visual data. E-commerce marketers routinely embed text inside images and skip alt tags entirely, which means a text extractor misses a huge chunk of what's actually on the page. ShopVision takes high-resolution screenshots and runs visual inference on top of them instead, which means they needed browser agents that could keep up.

Building and maintaining that browser infrastructure in-house would have been costly, and it would have pulled ShopVision's engineering team away from the work that actually makes their product unique: the AI models, the data pipelines, and the matching algorithms.

The solution: Browserbase powering every layer of the stack

ShopVision needed a browser agent platform that could render pages fully, handle access to any site with Agent Identity, and operate reliably at scale.

"Browser automation is not what makes us successful. I want my team to be experts in AI and e-commerce. I'd rather rely on the experts in browser automation to do the heavy lifting." — Jeff Neil, CTO

Marketing intelligence

Every day, Browserbase drives ShopVision's data collection of 150,000 e-commerce sites, gathering screenshots, paid social creative, and page-level data that feeds into their marketing intelligence models. It's the foundation their entire platform sits on.

ShopVision also uses Browserbase with Stagehand to programmatically sign up for competitor brand email and SMS lists, capturing every message brands send out as a live data source. Their browser agents don't just subscribe. They self-segment by preference, re-sign up periodically to account for email-age-based segmentation, and take actions on site (adding items to cart, for example) to get bucketed into different personalization flows. The result is a complete picture of how brands communicate with different customer segments over time.

Pricing intelligence

ShopVision's pricing intelligence product tracks how a brand is priced relative to its competitors, down to the variant level. The challenge is that some products have an astronomical number of configurations. A single business card SKU, for example, can have up to 19.1 million customization combinations. Trying to capture all of that in one browser session would take too long and almost certainly trigger detection. So instead, ShopVision uses a fan-out pattern: many small, short sessions running concurrently to cover every configuration they need while staying under detection thresholds.

Underpinning the whole thing is a fully autonomous pipeline built on Browserbase and Stagehand. When a new competitor site enters the pipeline, a catalog browser agent classifies it, then Browserbase spins up a session while Stagehand navigates the site and captures screenshots at each step. At the same time, a Claude Code agent writes a Playwright script to replicate exactly what Stagehand is doing. A second LLM then validates that script against the screenshots, and if anything doesn't match up, the loop runs again until the output is accurate. The result is a pipeline that writes, tests, and corrects its own collection scripts without any engineering intervention. That is what brought the pipeline down from 13 hours to 3.

Super Agent: from raw data to a browser agent your team can talk to

The Super Agent lets merchandising and marketing teams ask any business question in plain English and get answers backed by ShopVision's 1.2 billion data points combined with their own first-party data. Which SKUs overperformed last quarter? How did the top five competitors sequence their BFCM campaigns? Where are we missing homepage placements? Super Agent surfaces answers instantly, without SQL and without dashboards, delivered directly into Slack, Teams, or email.

The browser agents Browserbase enables on the data collection side are what make Super Agent possible on the intelligence side. Without reliable, real-time coverage of 150,000 sites, the answers would be guesswork.

The result: a massive and reliable competitive intelligence engine

ShopVision closed $1M+ in ARR in the first nine months of selling, entirely built on the data infrastructure Browserbase makes possible. They've run over 1.3 million browser sessions in the last 30 days and collected 1.2 billion data points since inception. That data represents a live, continuously updated index of e-commerce pricing, marketing activity, and product catalog changes across North America.

The drop from 13 to 3 hours in pipeline time tells a similar story. The fan-out architecture and autonomous script generation unlocked what ShopVision could actually deliver. Fresher data means customers are making pricing decisions based on what competitors are doing today, not yesterday. At the pace e-commerce moves, that gap matters. And because the architecture scales horizontally, ShopVision can take on more customers and more competitor catalogs without rebuilding how the pipeline works.

That faster, richer data layer also powers Super Agent directly. Every hour shaved off the pipeline means teams get competitive answers that are sharper, more current, and more actionable.

What's next

ShopVision is scaling fast. With growing demand for competitive market and pricing intelligence, their next phase is bringing those products to more customers at even higher concurrency. Long term, the vision is to move from intelligence to action: using their matching data and price history not just to inform decisions, but to eventually recommend and execute pricing changes directly. Super Agent is the interface that makes that future possible, and Browserbase is the infrastructure underneath it.

Interested in building browser agents at scale? Give your agents the web →