On February 25, 2026, Perplexity CEO Aravind Srinivas didn’t just ship a new feature — he shipped a different theory of what software should be. Perplexity Computer launched as a cloud-based multi-agent AI system that Perplexity describes as “a general-purpose digital worker.” It’s not a physical computer. It’s not a chatbot with extra steps. It’s closer to a persistent AI employee that orchestrates nearly 20 frontier models simultaneously, routes tasks to whichever model handles them best, and can keep working autonomously for hours or months at a stretch. Then, at the Ask 2026 developer conference on March 11, Perplexity went further — announcing Personal Computer (a 24/7 agent running on a dedicated Mac mini), enterprise integrations across Slack, Salesforce, and Snowflake, and Comet, an AI-native browser built specifically for organizations. This is the most coherent attempt yet to answer the question: what does software look like when AI is the operating layer, not the add-on?
What Perplexity Computer Actually Is (And Isn’t)
Let’s be precise, because the branding is genuinely confusing. Perplexity Computer is not a computer. There is no hardware in a box. It’s a cloud-based orchestration system that coordinates multiple AI models to execute complex, multi-step tasks on your behalf. The framing is intentional — Srinivas is positioning this not as a tool you use, but as a worker you assign tasks to.
The engine under the hood is what makes this technically interesting. Rather than betting everything on a single model, Perplexity Computer routes tasks across 19 to 20 frontier models depending on what each one does best:
- Claude Opus 4.6 — core reasoning engine for complex, multi-step logic
- Gemini — deep research tasks and spinning up sub-agents
- Nano Banana — image generation and processing
- Veo 3.1 — video creation and editing
- Grok — speed in lightweight, fast-turnaround tasks
- GPT-5.2 / ChatGPT 5.2 — long-context recall and broad search coverage
The architecture is deliberately model-agnostic. As better models arrive, Perplexity can swap them in without rebuilding the system. This is a real design advantage — rather than locking into one provider’s roadmap, Perplexity essentially bets on the AI field improving broadly and absorbs that improvement automatically. It connects to over 400 apps, and at current pricing sits behind the $200/month Max subscription tier.
The internal benchmark Perplexity released is worth sitting with: the system processed 16,000 queries over four weeks, performed what they equate to 3.25 years of human labor, and saved an estimated $1.6 million in labor costs. Those numbers are self-reported, so treat them as directionally interesting rather than independently verified — but the order of magnitude is consistent with what serious agentic deployments are showing across the industry.
Personal Computer: The Always-On Agent in Your Home Office
The announcement that got the most attention at Ask 2026 was Personal Computer — and for good reason. This one actually involves hardware. Perplexity ships you a dedicated Mac mini (M4, maxed out on RAM) that runs 24/7, connected to both your local applications and Perplexity’s cloud infrastructure. The agent has persistent access to your environment. It doesn’t forget context between sessions. It doesn’t go to sleep.
The practical implication is significant: you can assign long-horizon tasks — monitor this inbox, track these competitors, update this spreadsheet when new SEC filings drop, prepare a briefing every Monday at 7am — and the system executes them without you babysitting a prompt window. This is closer to what Andrej Karpathy has described as the shift from AI as a tool to AI as a colleague: something that has state, memory, and ongoing responsibility.
Perplexity has been thoughtful about the security surface here, which matters because the failure modes are real. Sensitive actions require explicit user approval before execution. There’s a full audit trail of everything the agent does. There’s a kill switch. This isn’t just good product design — it’s a direct response to the trust problem that’s slowed enterprise adoption of autonomous agents everywhere. People are willing to let AI act on their behalf, but only if they can see what it did and stop it when something goes wrong.
Personal Computer is waitlist-only as of March 2026, and requires the $200/month Max tier. Early access is limited, so the real-world performance data outside of Perplexity’s own benchmarks is still thin. Worth watching, but not worth reorganizing your workflow around until broader availability.
The Enterprise Play: Where Perplexity Is Actually Competing
The enterprise announcements at Ask 2026 are arguably more significant in the near term than Personal Computer, because they’re available now and they’re aimed squarely at where real budget decisions get made.
Computer for Enterprise ships with Slack integration — teams can @computer directly in channels and assign tasks without leaving their existing workflow. There are native connectors for Snowflake, Salesforce, and HubSpot, which means the agent isn’t just browsing the web — it’s reading and writing to the systems that actually run businesses. The compliance stack is serious: SOC 2 Type II certification and SAML SSO, the two checkboxes that tend to unlock procurement conversations at mid-size and large companies.
Developers get four APIs: Search, Agent, Embeddings, and Sandbox. That’s a meaningful surface area for building agentic workflows on top of. If Perplexity can establish itself as the infrastructure layer that developer teams build agentic workflows on, the platform dynamics get interesting fast.
Comet, the AI-native browser for organizations, is positioned specifically around knowledge work that involves financial data. It ships with over 40 financial data integrations — SEC filings, FactSet, S&P Global, Coinbase, and others. For analysts, investors, finance teams, and anyone whose job involves pulling data from multiple sources and synthesizing it, Comet is designed to make the browser itself the orchestration layer rather than a tab you switch away from.
Perplexity is currently valued at $20 billion. Their ARR was approximately $148 million at the midpoint of 2025, and they’re targeting $656 million by the end of 2026. That’s an aggressive growth target, and the enterprise push is clearly the mechanism they’re betting on to get there.
How Perplexity Stacks Up Against the Real Competition
Perplexity isn’t the only company building in this direction. Here’s an honest comparison of where things stand in March 2026:
| Product | Core Approach | Strengths | Weaknesses |
|---|---|---|---|
| Perplexity Computer / Comet | Multi-model orchestration, AI-native browser | Model-agnostic architecture, polished UX, strong
Recent PostsChatGPT Ads Hit $100 Million in Six Weeks: What OpenAI's Ad Platform Means for the AI Industry OpenAI launched ads inside ChatGPT and crossed $100 million in annualized revenue in six weeks. With self-serve access opening in April 2026 and projections reaching $100 billion by 2030, here's what... 74% of AI's Value Goes to 20% of Companies: What PwC's AI ROI Gap Study Reveals PwC's 2026 AI Performance Study shows 74% of AI's economic value is captured by just 20% of organizations. The top performers generate 7.2x more AI-driven gains — not through more AI, but through... |