Parag Agrawal’s Parallel Launches AI Platform That Outperforms GPT-5 in Online Research

AI Startup Parallel Web Systems Launches New Cloud Research Platform

Parallel Web Systems Inc., a two-year-old startup led by former Twitter CEO Parag Agrawal, has introduced a cloud platform designed to support artificial intelligence (AI) applications in conducting online research. Backed by investors such as Khosla Ventures, First Round Capital, and Index Ventures, the company recently raised $30 million to develop its technology further.

How Parallel’s Platform Works

The platform allows AI models to integrate data from public web sources directly into their prompt responses. It currently handles millions of research tasks daily for early users, including some of the fastest-growing AI companies.

Its uses are diverse. For instance, an AI programming assistant can retrieve relevant code snippets from GitHub repositories, while retailers might gather competitive product information from rival catalogs. The platform offers access to eight different AI research engines, each varying in speed and depth of data retrieval.

• Fastest engine: Processes requests in under one minute with cost-efficiency.
• Ultra8x engine: Provides detailed information but may take up to 30 minutes per task.

Each response includes confidence scores to help users evaluate data quality and citations for easy verification. Users can also request that gathered information be organized in specific formats, such as compiling product reviews into spreadsheets. Additionally, the platform offers customizable computing resources to optimize task completion.

Performance Compared to GPT-5

Before launching, Parallel tested its Ultra8x engine against GPT-5 using BrowseComp and DeepResearch Bench benchmarks. The results showed that Ultra8x outperformed GPT-5 by more than 10% in online research capabilities.

Access and Future Plans

AI models connect to Parallel’s platform through three APIs:

• Task API: General-purpose interface for various research needs.
• Search API: Optimized for powering AI agents.
• Chatbot API: Designed for low-latency applications like chatbots.

The company aims to broaden the platform’s use cases, hoping to enable AI agents capable of completing team-level workloads within hours. They are also developing features that allow agents to automatically respond to webpage changes.

This new platform offers practical tools for AI developers and businesses looking to enhance their models with real-time, verified web data. For those interested in further AI learning and applications, exploring specialized courses can be a valuable next step.

Explore AI courses and resources at Complete AI Training to deepen your understanding of AI applications and tools.