Nimble raises $47 million to give AI agents access to real-time web data

Believe it or not, web search is still thriving as an industry. As companies invest in using AI agents to get the most out of their data, there is a demand for tools that not only scrape the web to inform what these AI bots are doing, but also return those results in a way that is easier to use with modern data tools.

That’s the promise behind the web search startup Smartwhich recently raised a $47 million Series B round led by Norwest. The New York-based company’s platform uses AI agents to search the web in real time, verify and validate the results, then organize the information into ordered tables that can then be queried like a database.

That last part is important here. LLMs and AI agents are great for searching the web, correlating results from a variety of sources, and analyzing them, but they often display results in plain text, which can be difficult to work with at an enterprise level. And that’s before you take into account hallucinations, the risk of the client misunderstanding your instructions, or using unreliable sources.

By validating the results and organizing them into tables, Nimble allows companies to use web data as if it were already part of their existing databases. The startup also integrates with enterprise data warehouses and data lakes — large central repositories where companies store and analyze data — offered by the likes of Databricks and Snowflake. This means their AI agents can plug in a wealth of company-specific data, use it to build context, and shape how search results are organized and returned.

In effect, this allows organizations to have live, structured web data as part of their existing data environments, Nimble CEO and co-founder Uri Knorovich (pictured above, center) told TechCrunch.

These integrations also allow Nimble to remember constraints, such as how you want to search, or which data sources to click. This is especially useful for applications such as competitor analysis, pricing research, know your customer (KYC) processes, brand monitoring, deep research, and financial analysis. (Knorovich noted that Nimble works to ensure all customer data remains within its customer data infrastructure to comply with data retention and security policies.)

To that end, the startup has partnered with Databricks, Snowflake, AWS, and Microsoft to help simplify enterprise deployments that require access to internal data sources. (Databricks also participated in this Series B.)

“Models can do a lot of things, but most AI failures in production are not because the models aren’t good enough, but because the data fails,” Knorovich said. “What we’re seeing today is that organizations don’t need more AI; they need AI with good, reliable web search (…). If you can do that, if you can choose what your agent can search for and what they can’t search for, that’s the tipping point where organizations have to say, ‘Hey, we can actually trust AI.’ “We can actually put AI to work in more use cases.”

The ability to search the web in real-time and at scale, and to validate and organize search results, is what sets Nimble apart from other data brokers already in the space, says Knorovich.

The startup currently has over 100 clients, with the majority of its revenue coming from large enterprises, Fortune 500 companies, and even some Fortune 10 companies, including major retailers, hedge funds, banks, and consumer packaged goods companies, as well as some AI-driven startups.

“Nimble addresses a problem that has existed for years without a proper solution and is now of critical importance,” Assaf Harel, a partner at Norwest, said in a statement. “Reliable live web data is increasingly becoming a prerequisite for AI agents making critical business decisions.”

The Series B also saw participation from returning investors such as Target Global, Square Peg, Hetz Ventures, Slow Ventures, R-Squared Ventures, J-Ventures and InvestInData. Proceeds from the round will be used to expand research and development into multi-agent web search and a controlled data layer that processes and validates search results.

Nimble has now raised a total of $75 million.