Skip to main content
TechnologyJul 8, 2026· 2 min read

Perplexity Challenges OpenAI and Anthropic: 'Teammate', the New AI Tool for Developers, on the Way

Perplexity is preparing to enter the competitive market of AI-powered programming tools with a new internal project called Teammate. According to Business Insider, the San Francisco company, known primarily for its AI-driven search engine, has developed a platform dedicated to developers that may be made publicly available in the future, although there is currently no confirmation regarding a launch date.

The software has been used internally by the company's engineers since last May. Some screenshots obtained by the outlet show that Teammate is already being utilized for tasks such as identifying errors in internal systems and supporting more complex software projects.

According to available information, Teammate has been designed to follow the entire development lifecycle of a project. The goal is not limited to code generation but to assist developers in managing long-term tasks, analyzing problems, and monitoring services throughout the development process.

Further Details on Teammate: Perplexity's New Project

One of the highlighted features concerns the model-agnostic nature of the tool. This means that the system does not depend on a single AI model or specific chatbot, offering greater flexibility in choosing technologies to use.

A potential public launch would position Perplexity to compete directly with companies that are already very active in the AI programming tools sector, such as Cursor, Anthropic, and OpenAI, which have strengthened their presence in recent months with solutions dedicated to developers.

Business Insider also reports that Perplexity's CTO, Denis Yarats, strongly encourages the use of AI during software development. In some internal messages, the executive has stated that by the end of the year, or even sooner, engineers should increasingly rely on AI instead of working directly on code. Yarats has also dismissed criticisms that automatic generation tools produce low-quality code. In his view, the issue does not arise if the final result passes the appropriate quality checks established during development.