Perplexity Computer Brings Agentic Workflows Into Microsoft Teams Threads
Perplexity’s Academy tutorial presents Computer for Microsoft Teams as an AI agent meant to run inside Teams conversations rather than in a separate Perplexity interface. The company argues that users can install Computer from the Teams marketplace, use it in direct messages for private or early-stage work, and tag it in shared channels when teammates need visibility or context. Its broader claim is that agentic workflows — research, analysis, dashboards, reports, presentations, apps and websites — can be initiated, clarified and revised in the same threads where teams already coordinate work.
Perplexity is moving agent work into Teams threads
Perplexity presents Computer for Microsoft Teams as an AI agent that operates inside the collaboration layer where an organization already works. The operational point is not just access. It is that requests, clarifying questions, follow-ups, and outputs can live in Teams threads instead of requiring users to switch to the Perplexity app on desktop or phone.
The setup path is straightforward. A user goes to the Teams App Marketplace, searches for “Perplexity Computer,” and adds it, or asks a Teams administrator to install it. Perplexity also provides a Teams installation page for Computer.† The first time Computer is queried, Teams asks the user to sign in. The authorization flow then asks the user to review permissions, select “Allow,” and complete setup between Perplexity and Microsoft Teams.
That setup matters because the product is designed to work from inside existing workspaces and, when authorized, draw from connected tools. If a task requires access to a tool the user has not connected, Computer prompts for that connection before proceeding. In the demonstrated workflow, a Teams prompt asks the user to connect Salesforce before Computer can complete the requested task.
Private messages and shared channels support different kinds of work
Computer has two interaction modes in Teams: a direct message, or a channel or group message. Perplexity distinguishes them by the visibility and collaboration required.
Direct messages are for individual projects, early drafts, private research, and work that does not belong in a shared channel. The interaction is described as similar to messaging a colleague: type a prompt in natural language and send it. If Computer needs a connector that is not yet authorized, it asks the user to connect that tool, then proceeds with the task after authorization.
Computer may also ask clarifying questions before it starts. In the demonstrated dashboard task, it asks multiple-choice questions about dashboard format and ranking windows. Perplexity’s explanation is that answering these questions helps Computer lock onto the goal, scope, and expected output. The resulting example is a generated web dashboard for HelixNova Biopharma with an executive summary, metrics, and a “Top 5 Accounts” snapshot.
Shared work happens in a Teams channel or group message. A user opens the relevant space, tags @Computer, and describes the task in plain language. Perplexity says this mode is useful when teammates need visibility or want to contribute context. In the channel example, Computer replies in-thread with a plan and deliverables, may ask clarifying questions, and then begins working independently.
The thread remains the control surface for revision. A user can refine the prompt, change the requested output, or ask for additional deliverables. One demonstrated thread shows a user refining the output and Computer returning downloadable PDF briefs.
The Teams version is presented as carrying the same task scope as the web app
Perplexity says Computer can handle the same work in Teams that it can handle in the web app. The examples span web search, retrieval from connected organizational tools, content creation, business-data analysis, dashboards, reports, presentations, apps, and websites.
The visual examples make the promised scope concrete. Computer is shown producing a dark-themed AdPulse daily performance dashboard with spend, subscriptions, and a bar chart. It is also shown generating a report titled “Lumen Threads Quarterly Business Review,” and a daily report summarizing changes across connected tools such as Figma, Notion, and Jira. Another collage shows generated outputs including a chart, a document, a code interface, and a web page.
The key product claim is that Teams changes the operating context rather than the underlying task list. The same agentic workflows can be started privately in a DM or publicly in a channel thread. In a DM, the work can stay exploratory or contained. In a channel, teammates can see the prompt, add context, observe clarifying questions, and review the resulting output where the project discussion is already happening.
A public Computer channel is proposed as an adoption mechanism
Perplexity’s final recommendation is organizational rather than technical: create a public Teams channel where people can run Computer tasks. The stated purpose is to build familiarity with AI across the organization and let employees learn from one another by observing how AI is being used.
The demonstrated “Computer-Tasks” channel shows multiple users submitting requests and generating reports. In that setup, AI use becomes visible as workplace practice rather than a set of isolated experiments. Team members can see how prompts are written, how Computer asks for clarification, how users refine outputs, and what kinds of artifacts come back.
That recommendation clarifies the intended role of Computer in Teams. It is not presented merely as a chatbot embedded in another interface. It is positioned as an AI coworker that can be messaged privately, summoned into shared channels, connected to organizational tools, asked to clarify ambiguous requests, and refined through the same conversational threads the team already uses.