Aravind Srinivas, co-founder and CEO of Perplexity AI, has addressed user concerns regarding recent product updates and performance issues.

In a Reddit post, he explained the changes, outlined plans for improvement, and assured users of the company’s long-term stability.

Srinivas discussed the introduction of “Auto” mode, which aims to simplify the user experience by automatically selecting the most suitable AI model for each query.

He noted that this feature is still being refined. Manual model selection has become complex with the increasing number of options.

Users have reported inconsistencies across modes and the removal of the model selector in settings.

Srinivas clarified that specific models are designed for particular tasks. The Deep Research mode uses a combination of models, making manual selection unnecessary. He added that Perplexity plans to merge the Pro and Reasoning modes into one option for consistency.

Srinivas highlighted infrastructure challenges as a reason for recent issues, including user interface bugs and reduced reliability.

To address these problems, the company is prioritizing infrastructure improvements over new features. This includes transitioning from Python to GoLang and developing a more powerful deep research agent.

In response to feedback about follow-up queries defaulting to Auto mode, Srinivas stated that this change aims to improve efficiency.

Data showed users often skipped lengthy responses. The company plans to make certain models “sticky” to retain user preferences during follow-ups.

Srinivas also addressed speculation about the company’s financial health and potential IPO plans.

He affirmed that Perplexity is financially stable, with increasing revenue and sufficient funding. He clarified that there are no IPO plans before 2028.

Recent Perplexity developments