Small publishers lost 60% of search referrals. Google beat a publishers antitrust suit. For brands, the game is becoming a source inside the answer layer.
Perplexity Ships Computer and Memory Upgrades in Push Beyond Search
Perplexity launched Computer — a 19-model orchestration system priced at $200/month — and pushed memory recall to 95%. The company is pivoting from search to full agentic AI.
Perplexity is no longer just an AI search engine. On February 27, 2026, the company shipped Computer — a unified workspace that orchestrates roughly 19 different AI models to plan, delegate, and complete complex workflows end to end. Three weeks earlier, it rolled out major memory engine upgrades that pushed recall accuracy to 95%. Combined with a deep Samsung Galaxy S26 integration, these moves signal a deliberate pivot from search-centric AI toward a full agentic platform.
What Computer Actually Does
Computer is not a chatbot with extra steps. It is a multi-model orchestration system designed to function as what Perplexity calls a "general-purpose digital worker." You give it a high-level objective. It decomposes that objective into subtasks, selects the best AI model for each one, and executes the full workflow — potentially running for hours without requiring additional user input.
The model roster is specific. According to TechCrunch, Anthropic's Claude Opus 4.6 handles orchestration logic and coding tasks. Google's Gemini powers deep research queries. Google's Nano Banana generates images, and Veo 3.1 handles video. xAI's Grok handles lightweight, speed-sensitive tasks. OpenAI's GPT-5.2 manages long-context recall and expansive web search.
The core bet is that specialized models coordinated together outperform any single general-purpose model. Rather than building one model that does everything adequately, Perplexity is routing tasks to whichever model does that specific task best. It is an aggregation strategy — the AI equivalent of hiring specialists instead of generalists.
Pricing and Access
Computer is available exclusively on Perplexity's Max tier at $200 per month. Max subscribers receive 10,000 monthly credits plus a one-time 20,000-credit launch bonus that expires after 30 days. Pro and Enterprise access will follow after initial load testing.
At $200 per month, this is not priced for casual users. It is positioned for professionals and power users running sustained workflows — research, content production, data analysis, project coordination. The pricing reflects the compute cost of running 19 models in parallel, but it also reflects Perplexity's bet that users will pay substantially more for an AI that can complete complex tasks autonomously.
Memory Hits 95% Recall
Three weeks before Computer launched, Perplexity shipped a significant memory engine upgrade on February 6. The system now recalls relevant past interactions with approximately 95% accuracy, up from 77% previously — while making half as many memories. The shift is from volume to precision: fewer stored memories, but far more accurate retrieval.
Memory now extends to the Model Council feature, meaning personalized context follows across models. When Perplexity routes your query to a specific model, relevant memory travels with it. This matters for Computer's multi-model architecture — without reliable memory, a system that delegates across 19 models would lose coherence between tasks.
The February 6 release also included improved browser agent capabilities and expanded enterprise security controls, including SOC 2 compliance features. These are table-stakes for enterprise adoption, which Perplexity clearly has in its sights.
Samsung Galaxy S26: System-Level Integration
Perplexity's partnership with Samsung puts the assistant at the operating system level on the Galaxy S26. Users can invoke it with "Hey Plex" or by holding the side button. This is the first time Samsung has granted system-level OS access to an app that is not made by Samsung or Google.
The integration is deep. Perplexity can read from and write to Samsung Notes, Calendar, Gallery, Reminder, and Clock, as well as select third-party apps. It is not a standalone app sitting on top of the OS — it is woven into the device's workflow layer.
Samsung's framing is a "multi-agent ecosystem" where Galaxy AI orchestrates between Perplexity, Gemini, and Bixby based on task context. Perplexity's APIs also power complex queries through Bixby across roughly 800 million Samsung devices. That is distribution at a scale Perplexity could not achieve alone.
The Agentic Pivot
Zoom out and the pattern is clear. Perplexity is building three things simultaneously:
Persistent memory — 95% recall accuracy across sessions and across models
Device-level integration — system-level access on Samsung's flagship hardware
These are the three ingredients for an agentic AI system — one that remembers context, calls on the right tools, and integrates deeply into devices and apps. This is a different product category than search. Search answers questions. Agents complete tasks.
The competitive implications are significant. OpenAI, Google, and Anthropic are all moving toward agentic capabilities, but none of them currently offer a product that orchestrates competitor models together. Perplexity's willingness to use Claude, GPT, Gemini, and Grok in a single system is a differentiator — though it also creates dependency on providers who may not remain cooperative.
What This Means for Publishers and Brands
For anyone in the business of creating content that AI systems surface, Perplexity's evolution matters for two reasons.
First, agentic systems consume content differently than search engines. A system running a multi-hour research workflow will access, synthesize, and cite far more sources than a single search query. That is both an opportunity (more citation surface area) and a challenge (your content needs to be machine-parseable enough to be useful in automated pipelines).
Second, memory changes the relationship between brands and AI systems. If Perplexity remembers a user's preferences, industry, and past queries, the sources it surfaces will be filtered through that context. Brand visibility in AI is not just about being cited once — it is about being reliably present in the context window when relevant queries arise.
Perplexity is betting that the future of AI is not a single model that answers questions, but a coordinated system that completes work. Whether that bet pays off at $200 per month remains to be seen. But the direction — from search to agency, from answers to outcomes — is one the entire industry is moving toward.
James Calder is the editor of The Search Signal, covering AI-powered search, generative engine optimization, and the future of brand discovery.
Small publishers lost 60% of search referrals. Google beat a publishers antitrust suit. For brands, the game is becoming a source inside the answer layer.
Perplexity's Personal Computer turns a Mac mini into a persistent, always-on AI agent that works across your local files and apps around the clock — a new category of ambient computing.
