
/Customer4 min read
How Rox Cut Account Research from a Week to Seconds with Tavily
Millions of accounts. Fresh signals in seconds. See how Tavily helps Rox turn the live web into revenue intelligence at enterprise scale.
About Rox
Rox builds revenue agents for the Global 2,000, including MongoDB, Ramp, and New Relic. Founded in February 2024, its agents work the entire revenue pipeline: inbound, outbound, deal management, and renewals. Walk into a deal with Rox running underneath you, and you already know what happened at the account before you dial: the new CRO, the funding round, the product launch from that morning. An agent did the digging before you had your coffee.
The challenge
Accounts don't sit still. Leadership turns over, funding lands, products ship, often in the same week a rep is heads down running their pipeline. For Rox's agents to be worth trusting, they had to work off what's happening right now, not what was true last quarter. That meant pulling structured data straight out of the CRM and warehouse, unstructured records like emails and call notes, and whatever was moving on the open web, and stitching all of it together. Reps saw it land on their own schedule: a Monday morning catch-up on everything that moved across their accounts that week, plus a nightly pass that picked up whatever happened since yesterday.
Whatever sat underneath that system had to be clean enough for a reasoning agent to be trusted across thousands of accounts, and fast enough to answer while a rep is waiting, all at once. Plenty of tools handle one of those well. Almost none handle the whole set together. Rox also wanted its own engineers heads down on the reasoning layer that actually makes the product different, not nursing a web scraper back to health every other month. Get the web layer wrong, and a rep loses hours per account, and close to a week across a full territory: exactly the tax the agents exist to erase.
Sales is inherently a game which is constantly moving. Sales reps are actually looked at by how fast they can respond to the information that is happening on the internet and how quickly they reach out. Both freshness and speed are extremely important from a rep's perspective.
Shriram Sridharan, Co-founder & CTO, Rox
Why Tavily
Rox got here by testing, not guessing. They started with an alternative search provider on the market. Costs got out of hand fast, and customers kept flagging results that were stale. They tried another provider that paired search with its own reasoning layer. The search itself was affordable, but the reasoning tokens stacked on top made the total cost pricier.
A personal introduction brought Rox to Tavily. Rotem, Tavily's CEO, worked the implementation himself. When Rox needed date range filtering that didn't exist yet, Tavily built it in a day or two and handed over a live endpoint to test right away, no roadmap promise, no "next quarter." Rox ran its own evals, starting with live production prompts its engineers validated manually and now run through a fully automated pipeline. Tavily came out ahead on freshness, accuracy, and price per million queries. The API was live on the first call, and Rox's production workload had moved over within two weeks.
I haven’t seen this in any other providers that we've worked with... The team is just phenomenal to work with, and in this age of things that are fast moving, we just want to partner with somebody who is also at the same pace, right from the CEO to everybody on the team.
Shriram Sridharan, Co-founder & CTO, Rox
How Rox uses Tavily
Tavily handles search and extraction. Rox handles everything on top, ranking, deduping, reasoning, so the output looks exactly the way its agents expect it. Clever Columns pulls fresh company news into any row the moment a rep asks for it. Autopilot, Rox's agentic layer, uses Tavily to power account intelligence and personalized outbound at scale. Tavily also sits behind Rox's chat interface, and quietly backs up the rest of the platform whenever it needs the open web.
The scale is real: millions of Tavily requests a day, each one answered in milliseconds. Stack Rox's own ranking and reasoning on top of that, and a fresh signal lands on a rep's screen in a few seconds, about as long as it takes to glance at your phone.
The results and what’s next
Rox now tracks millions of accounts and billions of contacts on this infrastructure, the same infrastructure it stood up in two weeks flat. Research that used to eat a rep's week now runs itself overnight and is simply waiting there before anyone thinks to ask for it. Rox has grown fast this year and they credit nearly all of that growth to what Tavily powers underneath. Just as important, its engineers get to spend their time on the reasoning layer that actually makes the product different, instead of keeping a search stack alive. Rox's next move is to push deeper into the bottom of the funnel, into areas like CPQ and revenue recognition, beyond the top and mid funnel work it already owns. Every new account or contact it brings on needs that same web layer underneath. Its scale keeps making Tavily's search sharper, and every improvement lands right back in the product.
With Tavily, our agents can confidently rely on a web search provider that’s fast, cost effective and provides accurate data at scale.
Shriram Sridharan, Co-founder & CTO, Rox
Ready to turn real-time web data into revenue? We’re ready when you are. Contact sales here.
