Former Twitter CEO Parag Agrawal’s AI Startup Parallel Beats GPT-5 in Deep Web Research

Nearly three years after his abrupt departure from Twitter (Now X) following Elon Musk’s takeover, Parag Agrawal has launched a new AI startup, Parallel Web Systems, where he serves as founder and CEO. 

Backed by prominent venture capital players, including Khosla Ventures, Index Ventures, and First Round Capital, the startup has raised approximately $30 million in funding.

Parallel’s founding team includes engineers who previously worked at Twitter, Google, Stripe, Airbnb, Chime, Waymo, and Kitty Hawk.

What Is Parallel Building

Parallel aims to empower AI systems with the ability to interact with the public web in real time. Instead of relying on static data, the system enables AI agents to fetch, verify, organise, and even assess the confidence of the information they present, essentially giving AI its own browser.

“Agents are going to be the primary customers of the web going forward. They will use the web a lot more than humans ever have,” said Agrawal.

Parallel has also released its deep research APIs that outperform both humans and leading AI models, including GPT-5, on benchmarks testing complex web search and research tasks.

The company reported that on OpenAI’s BrowseComp benchmark, which measures accuracy in multi-hop reasoning and web navigation, Parallel achieved 58% accuracy compared to 41% for GPT-5 and 25%  for human participants working with two-hour time limits. 

On DeepResearch Bench, which evaluates long-form synthesis across 22 fields, Parallel recorded an 82% win rate against a reference set, compared to 66% for GPT-5.

“AIs will use the web far more than humans ever have. As much as we might anthropomorphise AIs, they operate differently from humans on the web,” the company said in its announcement. “At Parallel, we are building for the web’s second user. We are creating systems and infrastructure for AIs to use the web effectively for completing complex tasks. ”

Parallel said its APIs are already powering “millions of research tasks daily” for startups, public enterprises, and AI companies. According to the company, coding agents use the platform to search documentation and debug, while enterprises deploy it to automate workflows such as insurance claims processing.

Parallel’s infrastructure was built specifically for AI agents, with each layer of its stack, from crawl and index to ranking, engineered for machine consumption rather than human browsing. The company said this approach allows “declarative interfaces” where agents specify what they need, without detailing how to retrieve it.

Looking ahead, Parallel said it is developing capabilities for long-horizon agents, continuous monitoring of web signals, event-driven architectures, and SQL-style programmable queries over the web. “We’re building for what’s next,” the company said.

Agrawal’s tenure at Twitter ended in 2022 after Elon Musk’s $44 billion takeover. With a background spanning IIT Bombay, Stanford, and a decade at Twitter, he is now focusing on his first major project since then, building Parallel Web Systems to advance AI infrastructure.

The post Former Twitter CEO Parag Agrawal’s AI Startup Parallel Beats GPT-5 in Deep Web Research appeared first on Analytics India Magazine.