-

Firecrawl Shows How Enterprises Can Close the Agentic AI Gap

Five steps help companies move from asking AI questions to assigning agents whole jobs, backed by official sources

SAN FRANCISCO--(BUSINESS WIRE)--Firecrawl, the company that gives AI agents access to the world’s knowledge through a single API, today published AI’s Search-Bar Problem, a white paper on why most companies aren’t yet seeing bottom-line value from agentic AI, and five steps to close that gap.  

Early television didn’t reach its potential until it stopped putting radio shows in front of a camera. AI is at the same stage: most of its value lies in doing work, not answering questions. Yet most people still use it like a search bar. The largest public study of consumer ChatGPT use found that about 49% of messages ask for information or advice, 40% ask the model to produce something and 11% express a view. Enterprise research shows a similar gap: McKinsey reports that nearly eight in ten companies use generative AI, yet just as many report no significant bottom-line impact.

The paper highlights five ways to overcome this gap: 

  1. Measure how much of a team's AI use is asking rather than assigning 
  2. Write jobs instead of prompts 
  3. Map the sources each job depends on 
  4. Require a source link for every fact an agent acts on
  5. Score finished work rather than answers 

“Most people still type a few words into AI, take the first answer and close the tab, because that’s what twenty-five years of search taught us to do,” said Laura Franzese, head of brand at Firecrawl. “That habit works for finding a web page. It doesn’t work for something that can read, compare, check and finish the job. The companies seeing results are the ones handing agents whole jobs, not questions.”

The root of the gap is simple: an agent’s work is only as good as the information it can reach. Stanford researchers found that general-purpose chatbots hallucinated on 58% to 82% of specific legal questions. Firecrawl’s own testing points the same way. Firecrawl gave an AI agent 581 requests for official US legal sources, such as statutes, regulations and court opinions, modeled on real customer queries. Working from memory alone, it put the official document first 47.8% of the time. Firecrawl’s Government & Legal Index, built from official sources, did so 76.2% of the time, 1.6 times as often.

“Official sources and receipts completely change what you can do with the output,” said Caleb Peffer, CEO and co-founder of Firecrawl. “You can check it, cite it and act on it. Getting agentic AI to the bottom line, instead of turning it into another expensive bottleneck, comes down to four habits: write jobs for agents, map their sources, require receipts for every action, and score the finished work.”

Firecrawl's Government & Legal Index covers more than 30 million data points of US law, and its Research Index gives agents more than 41 million research papers. In a separate test across 845 tasks, Firecrawl gave AI agents the same model and the same instructions, and changed one thing: where they got their information. Agents that could pull data through Firecrawl Alexandria gave answers that scored 21% higher than agents limited to built-in web search.

Availability

AI’s Search-Bar Problem is available free today at https://www.firecrawl.dev/whitepapers/ais-search-bar-problem

About Firecrawl

Firecrawl gives AI agents access to the world’s knowledge through a single API. Agents can search, scrape and retrieve from the live web, Firecrawl’s own indexes and Alexandria, its library of premium data providers and specialized indexes. More than 1.5 million users and 150,000 companies build with Firecrawl, including teams at 67 of the Fortune 100. Firecrawl pays its data providers and is building ways for more contributors to earn from the knowledge AI agents use. The company is based in San Francisco, with an office in Toronto. Learn more at firecrawl.dev.

Contacts

Media contact
press@firecrawl.com

More News From Firecrawl

Firecrawl Brings Its Library for the World’s Knowledge to Toronto

TORONTO--(BUSINESS WIRE)--Firecrawl, the company that gives AI agents access to the world’s knowledge through a single API, today announced the opening of its Toronto office, its first outside San Francisco. The new office, located downtown in the Entertainment District, will be home to engineering and product teams and will serve as the base of the company’s newly established Ontario subsidiary. AI agents now research companies, review scientific literature and write production code. How well...
Back to Newsroom