Perplexity Launches Computer for Enterprise to Rival Microsoft

Perplexity unveiled Computer for Enterprise at its inaugural Ask 2026 developer conference this week, transforming the search-startup-turned-AI-agent-platform into a direct competitor to Microsoft Copilot and Salesforce's Agentforce. The launch arrives just two weeks after Perplexity's Computer agent debuted for consumers, where viral demos of the product building Bloomberg-style financial dashboards and replacing six-figure marketing tool stacks pulled in a wave of enterprise inbound.

The pitch to CIOs is simple: one agent that orchestrates 20 different AI models, plugs into the tools your company already uses, and lives inside Slack where employees already work.

What Computer for Enterprise Includes

The product centers on a new Slack integration. Employees can now query @computer directly inside any Slack channel or thread, then continue the same conversation in Perplexity's web interface or mobile app — with full context, model routing, and connector access carried across surfaces.

On the backend, Computer for Enterprise ships with native connectors for Snowflake, Salesforce, and HubSpot, plus single sign-on, audit logs, and the data-retention controls enterprise security teams expect. Perplexity says no enterprise data is used to train its models.

There's also a desktop component. The Comet browser, Perplexity's AI-native browser, can now be deployed silently across macOS and Windows fleets via mobile device management. Administrators can configure hundreds of browser policies and lock down exactly which actions the embedded AI agent is allowed to take.

Why this matters: enterprise buyers have struggled to deploy generative AI safely at scale. Perplexity is bundling the agent and the browser together so security teams can govern both surfaces from one console.

The Multi-Model Bet

Computer's defining technical choice is that it runs across roughly 20 different AI models simultaneously, routing each request to the model best suited for the task — fast lightweight models for short queries, frontier models like GPT-5.5 and Claude Opus 4.7 for complex reasoning, and specialized models for code, search, or vision.

Perplexity calls this multi-model orchestration. The bet is that no single frontier lab will dominate every capability, and that the winning enterprise platform is the one that picks the right model for each step. That's a meaningful contrast to Microsoft Copilot (which leans heavily on OpenAI) and Google's Gemini Enterprise platform (which uses Gemini almost exclusively).

For customers, the practical benefit is cost and latency. Routing simple questions to cheap small models and only escalating to flagship models when needed can cut per-task costs significantly versus running everything through a single frontier model.

Enterprise Traction Signal

Perplexity claims more than 100 enterprise prospects messaged the company over a single weekend after the consumer Computer launch, demanding access. The enterprise version is rolling out with named launch customers, though Perplexity has not yet published a full customer list publicly.

The company's most recent reported valuation is $20 billion, raised in a round earlier this year. Annual recurring revenue figures have not been disclosed, but Perplexity has been growing search and chat usage rapidly across consumer and pro tiers.

Why this matters: Microsoft and Salesforce both have multi-year head starts, deep distribution into IT, and existing enterprise relationships. For Perplexity to compete, product velocity has to translate into a willingness to migrate workflows away from incumbents — and the early signal is that some companies are willing.

Industry Implications

For Microsoft, Computer for Enterprise targets the exact use cases Microsoft 365 Copilot was built to win: a knowledge worker asking a question in Slack or Teams and getting an answer that pulls from CRM, data warehouse, and the open web. Microsoft has spent the past two years refining Copilot's connector story; Perplexity has built a comparable connector library in a fraction of the time.

For Salesforce, the threat is more pointed. Perplexity Computer can read from Salesforce, generate dashboards, and draft emails — many of the same tasks Agentforce was designed for. Salesforce has the advantage of being inside the system of record, but Perplexity has the advantage of being model-agnostic and faster-shipping.

For OpenAI's recently launched Workspace Agents, Computer for Enterprise is the most obvious head-to-head competitor. Both products promise no-code agent building. The differentiator will come down to integration depth, governance controls, and which model orchestration story enterprise buyers find more credible.

Expert Perspectives

Industry analysts on X noted that Perplexity's enterprise launch coincides with a broader shift: agents are no longer research demos, they're being procured. Several CIOs cited the Slack-native experience as the key unlock — meeting employees where they already collaborate.

Critics pointed out that Perplexity's revenue base is still small relative to the incumbents it's targeting, and enterprise sales cycles are long.

What's Next

Computer for Enterprise is generally available now to enterprise contract customers. Pricing is consumption-based, with model usage and connector calls metered separately from seat licenses.

Expect Perplexity to roll out additional connectors over the next few months — likely Workday, ServiceNow, Notion, and Asana. The Comet browser is also expected to gain deeper Windows-fleet management features.

The bottom line: Perplexity has gone from search startup to enterprise AI platform faster than expected, and Computer for Enterprise is the most direct shot yet at Microsoft's and Salesforce's AI agent franchises. If you're evaluating an enterprise AI agent in 2026, this product now belongs on the shortlist.