Perplexity is discreetly developing a coding tool powered by AI.

      Perplexity aims to join the AI coding surge. The San Francisco-based search startup has developed an internal coding tool, known as “Teammate,” and might release it publicly in the future, according to a source who spoke to Business Insider. This tool has been utilized by its engineers since May.

      The company initially gained attention with its AI search engine that competes with Google. Introducing a coding product would shift its focus significantly, aligning it more with Cursor, Anthropic, and OpenAI, all of which already have popular coding tools in operation.

      What “Teammate” does

      According to an internal announcement reviewed by Business Insider, Teammate is designed to manage software projects from start to finish. “It’s intended for extensive engineering tasks: managing projects, troubleshooting problems, and overseeing services,” the message states.

      Screenshots reveal engineers assigning it real tasks, such as searching for bugs in internal systems. The tool remains model-agnostic, as mentioned by the source, meaning it does not rely on any specific chatbot, a deliberate choice in contrast to competitors like Claude Code that utilize their own models.

      The 💜 of EU tech

      Recent developments from the EU tech industry, a narrative from our knowledgeable founder Boris, and some questionable AI-generated art. It's free and delivered weekly to your inbox. Sign up now! A spokesperson for Perplexity declined to comment, and it is uncertain when the product will be launched. The startup was valued at $20 billion in a funding round last year, giving it the flexibility to explore new ventures.

      “Stop looking at code”

      The initiative is driven by leadership. A few weeks before Teammate was internally launched, Chief Technology Officer Denis Yarats instructed engineers that by the end of the year, or earlier, they should “stop looking at code” and simply utilize AI. This aligns with OpenAI's own internal transition to coding agents.

      Yarats also responded to claims that AI produces “slop,” or substandard code. “Slop will not be an issue,” he stated, as long as the generated code meets quality standards.

      Why it matters

      The timing increases the stakes. AI coding is one of the few areas where the technology is already generating substantial revenue, which explains the influx of well-funded players in the field. A model-agnostic tool from a $20 billion search company would introduce a new perspective to an already competitive landscape.

      However, two caveats should be noted. This information is based on a single source, and Perplexity has not officially confirmed a launch. Nonetheless, the trend is evident: even a company focused on AI search recognizes that developing your code, beyond merely providing answers, is a valuable opportunity.