Perplexity’s new Computer clones Bloomberg Terminal, runs multiple AI agents, draws

Perplexity’s “Computer” platform bundles a dozen‑plus large‑language models into a single, cloud‑hosted agent that can spin up subordinate agents to tackle discrete subtasks, a capability the company touts as a “unified AI computer” (TechCrunch). The service, which runs on the firm’s top‑tier Perplexity Max plan at $200 per month, claims to orchestrate 19 distinct models, allowing users to launch complex, multi‑step workflows without manually chaining prompts (TechCrunch). In demo‑style screenshots posted on Perplexity’s website, the tool pulls together financial statistics, legal references, and raw data, then assembles the results into polished visualizations or even full‑featured web pages, effectively acting as a one‑stop shop for research‑heavy tasks (TechCrunch).

The architecture mirrors the multi‑agent paradigm popularized by recent AI‑first products, but Perplexity differentiates itself by keeping the entire stack in the cloud, a design choice that “might spare it some of the security concerns of other agentic tools like OpenClaw” (TechCrunch). By offloading execution to remote servers, the platform avoids the need for users to install or manage local runtimes, a factor that could appeal to finance professionals accustomed to the high‑security environment of Bloomberg’s Terminal. Nonetheless, the move has sparked skepticism among technologists who question whether a single subscription can truly replace the depth of data, analytics, and compliance tools baked into Bloomberg’s $30,000‑a‑year service (Tom’s Hardware).

Industry observers note that Perplexity’s pricing sits at a fraction of Bloomberg’s annual fee, yet the value proposition hinges on the quality and freshness of the data sources the agents can access. While the company’s marketing material showcases impressive end‑to‑end examples, a scheduled press demo was abruptly canceled after “flaws found in the product hours before the event,” according to a background briefing report (TechCrunch). The cancellation underscores lingering reliability concerns and suggests the platform may still be in a polishing phase, despite the hype generated by its ambitious multi‑model claim (TechCrunch).

VentureBeat’s coverage emphasizes the strategic bet behind the launch: Perplexity is positioning the Computer as a proof point that “users need many AI models” to handle the breadth of modern enterprise queries (VentureBeat). By packaging 19 models under a single interface, the firm hopes to lock in high‑value customers who require both breadth and depth in AI‑driven analysis, from market research to legal compliance. The move also aligns with a broader trend of AI vendors bundling disparate capabilities—such as retrieval‑augmented generation, code execution, and specialized domain models—into unified agents, a pattern that analysts predict will intensify as enterprises seek to reduce the friction of juggling multiple subscriptions (Computerworld).

Critics, however, caution that the Bloomberg Terminal’s value lies not only in its data feeds but also in its extensive network of real‑time market information, proprietary analytics, and regulatory safeguards—assets that a cloud‑only AI orchestrator cannot instantly replicate (Tom’s Hardware). The skepticism is echoed in the finance‑tech community, where early adopters have praised the novelty of Perplexity’s approach but remain wary of its readiness for mission‑critical use cases. As Perplexity refines the Computer’s reliability and expands its data integrations, the platform may carve out a niche as a cost‑effective adjunct for analysts, but whether it can truly rival Bloomberg’s entrenched ecosystem remains an open question.