Perplexity Computer integrates Snowflake with SQL, source tables, and live metrics

• perplexity_ai's launch post says Perplexity Computer now connects directly to Snowflake and can answer questions against live warehouse data with SQL, source tables, filters, and metrics.
• In the same announcement, perplexity_ai's follow-up adds that Computer can turn those warehouse queries into dashboards and automations for pipeline analysis, product usage, and customer segmentation.
• perplexity_ai's follow-up also says admins keep control over access, business definitions, and shared data logic, which puts the feature inside governed analytics workflows instead of ad hoc chat over exports.
• AravSrinivas framed the integration as a way to set up "a fleet of on-call data scientist agents," while WesRoth's demo summary shows the product generating SQL and surfacing analysis results inside the Computer interface.

A week after Perplexity introduced Personal Computer, the company added a official blog post for a more enterprise-shaped use case. You can watch the launch demo query Snowflake from inside Computer, skim the follow-up post for the dashboard and automation angle, and see WesRoth's screen recording walk through generated SQL and result views.

Perplexity's core claim is narrow and useful: Computer can now connect to Snowflake and run end-to-end analysis on live warehouse data. According to perplexity_ai's launch post, the answers are returned with the SQL query, source tables, filters, and metrics attached, which is a lot more auditable than a plain natural-language summary.

Aravind Srinivas described that setup in agent terms, with AravSrinivas calling it "a fleet of on-call data scientist agents." The official framing in the official blog post stays closer to analytics plumbing than agent theater.

The governance detail is the part that makes this feel enterprise-ready. In perplexity_ai's follow-up, Perplexity says admins still control three things:

• access
• business definitions
• shared data logic

That means the integration is not pitched as a side door around warehouse policy. The company is explicitly tying Computer to the existing layer of permissions and metric definitions described in the official blog post.

Perplexity is also stretching Computer past question answering. perplexity_ai's follow-up says the Snowflake connection can be used to build dashboards and automations, with examples including pipeline analysis, product usage, and customer segments.

In WesRoth's screen recording summary, the flow goes from a natural-language prompt to generated SQL, then into a result view with options to build a dashboard or automation. testingcatalog's recap makes the same jump explicit, describing Computer as both a personal data science team and a data analytics team.