Perplexity’s newly announced “Perplexity Computer” has sparked intrigue across the Artificial Intelligence world. The system promises a future where autonomous multi‑agent workflows handle complex tasks end‑to‑end — browsing, reasoning, coding, summarising, and acting in sequence with minimal human supervision.
However, when three of today’s leading AI models — Google’s Gemini, Grok, and OpenAI’s ChatGPT — were asked by NDTV Profit to evaluate the concept, their reactions were far from unanimous applause. What emerges is a mix of admiration, skepticism, and pointed warnings about technical hurdles, pricing, and long-term viability.
Gemini: Turning AI Giants Into ‘Plumbing’
Google’s Gemini delivered the sharpest strategic warning: Perplexity’s system, if it scales, could effectively turn the world’s largest model-makers — Google, OpenAI, Anthropic — into “plumbing.” In other words, Perplexity becomes the user-facing layer while the big models are relegated to a commoditised backend.
To counter that future, Gemini predicts Google will embed agentic, multi-step workflows directly into Chrome and Android OS, making AI agents native to the world’s largest platforms. If true, Perplexity may find itself competing not just with other AI startups but with entire operating systems.
ALSO READ: Perplexity ‘Computer’ Just Dropped, And Your Operations Team Might Not Be Too Happy
Gemini also flags a deep technical problem: the “context window trap.” Passing large files, live browser data, and persistent memory across up to 19 different models without data loss is far from solved. Maintaining coherence across such a system, Gemini suggests, is one of the biggest unspoken challenges.
Grok: Useful, But ‘Not a Game-Changer’
Elon Musk’s Grok model takes a more pragmatic stance. It calls the Perplexity Computer impressive but “not a full game-changer”, largely because of today’s reliability ceiling. AI agents still hallucinate, get stuck in loops, or break down on edge cases. Sub‑agents may reduce failure rates, but Grok argues long-running autonomy will continue to require human oversight.
Then there’s the cost: at $200 a month plus usage-based credits, Grok says the product is built for professionals and enterprise workflows — not casual users or small teams. Heavy workloads could push costs significantly higher, limiting mass adoption for now.
ChatGPT: Power Meets Risk
ChatGPT echoes the concerns around autonomy, warning that multi-step agent workflows introduce compound-risk failures: small errors snowballing into large ones, integrations breaking, or agents performing unintended actions. As agentic systems become more capable, ChatGPT suggests, their failure modes become more unpredictable.
Pricing is another challenge. Early price points around $200/month position the Perplexity Computer as a premium tool. Meanwhile, the broader competitive landscape is heating up — from open-source agentic frameworks like OpenClaw to browser-based and workspace automation tools, and larger incumbents like OpenAI, Google and Microsoft building their own workflow engines.
Consensus And Bottomline From Competition
Despite their differences, all three models converge on four points:
(1) The Perplexity Computer is bold but not revolutionary.
(2) Autonomy still suffers from hallucinations and reliability gaps.
(3) Pricing will restrict adoption.
(4) Competition in agentic AI is about to intensify dramatically.
Perplexity’s Computer is one of the most ambitious attempts yet to package autonomous agents into a consumer-ready product. But when AI evaluates AI, the verdict is clear: the engineering is bold, the vision compelling — yet the system remains constrained by reliability limits, technical complexity, and steep pricing.
A step forward? Yes.
A revolution? Not yet.
ALSO READ: Coders To Custodians? Anand Mahindra Rejects Citrini’s AI Doom, But Warns Of Major Pivot
Essential Business Intelligence,
Continuous LIVE TV,
Sharp Market Insights,
Practical Personal Finance Advice and
Latest Stories — On NDTV Profit.
