Technical Context
I won't pretend to have seen the confirmed text of that specific tweet. Without a verified post, I'm focusing on Perplexity's official release trajectory, which is already quite transparent.
After digging into their latest updates, the picture is clear: Perplexity has long ceased to be just a search engine. They are building a layer for AI automation, where search, Deep Research, Computer, Workflows, MCP, and API begin to operate as a unified system.
What really caught my eye: MCP support enables connections to external tools and data, Computer moves the product toward managed agent execution, while Workflows and Skills hint at repeatable scenarios instead of one-off queries. In short, it's no longer 'ask a question, get an answer,' but the foundation of an environment where the model can search, decide, and act.
It's also interesting how Perplexity blends in multimodality, voice mode, enterprise integrations, and access to fresh models like the GPT lineup. To me, this signals AI integration at the stack level: data, orchestration, tools, and the human-in-the-loop, rather than just UI updates.
This is where I usually pause and ask a simple question: is this a toy or a functional layer? Judging by their focus on Teams, Snowflake, Databricks, and reusable workflows, they are clearly aiming for the latter.
Impact on Business and Automation
I see three practical shifts for businesses. First, search is no longer a separate tab; it becomes part of the operational loop. This means you can build AI solutions for business around analytics, research, compliance, and internal assistants much faster.
Second, architecture becomes far more valuable. If your data is scattered across CRMs, BI tools, spreadsheets, and internal databases, a new interface alone won't save you. You need a proper setup for access rights, tools, logging, and human-readable guardrails.
Third, teams that can quickly build agent scenarios on top of real processes will win. Those who fall for flashy demos without actual implementation will lose. I build these things hands-on, so I clearly see the difference between a 'wow' social media post and a system that saves hours every week.
If you have a pressing need to turn search, research, or internal operations into working AI automation, let's look at your process without the magic and marketing. At Nahornyi AI Lab, my team and I build these solutions for real workloads, ensuring Vadym Nahornyi won't have to explain later why a demo looked great but failed in business.