Chronological feed of everything captured from Aravind Srinivas.
The Comet AI browser assistant has launched on iOS, offering a modern, AI-powered browsing experience. Key features include conversational search, page summarization, tab management, and integration with Perplexity AI. This application aims to replace traditional browsers by incorporating AI assistance throughout the user experience, emphasizing design, fluid gestures, and real-time AI responses.
Comet is an AI-powered browser designed for enterprise use, focusing on security and manageability. It integrates with existing security infrastructure like CrowdStrike Falcon for threat detection and offers centralized control over agent operations. Deployment across large organizations is facilitated through Mobile Device Management (MDM).
Comet is an AI-powered browser designed for enterprise deployment, offering centralized management and integration with security platforms. It targets large-scale internal rollouts and provides granular control over agent operations.
A speculative concept proposes direct AGI integration via browser control, offering a tangible demonstration of AGI capabilities by observing its interaction with and manipulation of a user's digital interface. The central idea revolves around the AGI taking over the user's screen, specifically the pixel output, to illustrate its control.
A single individual, leveraging AI tools like ChatGPT and AlphaFold with a modest investment, successfully designed a custom mRNA cancer vaccine for a pet, achieving significant tumor reduction. This case demonstrates AI's potential to democratize and accelerate drug discovery, outpacing conventional pharmaceutical R&D workflows through rapid identification of mutated proteins and corresponding drug targets. The efficiency and accessibility of this approach suggest a paradigm shift in personalized medicine and disease treatment.
Perplexity AI, under CEO Aravind Srinivas, is evolving from an answer engine to a comprehensive AI assistant, prioritizing accuracy and contextual understanding. The company is focused on developing "Comet," a browser-based personal AI assistant, and asynchronous background agents to automate tasks and enhance user productivity. Srinivas emphasizes innovation, user-centricity, and a bias for action to stay competitive in the fast-paced AI landscape.
Perplexity AI aims to disrupt Google's search monopoly by prioritizing accurate, cited answers over ad-influenced results. The company's strategy involves a subscription-based model, building its own search index, and developing an agent-based browser to leverage AI for complex tasks. This approach targets the "knowledge discovery" market by fostering user curiosity and providing reliable information.
Aravind Srinivas posits that the next frontier of AI is not larger models, but the orchestration of end-to-end systems that handle user context and tool imperfections. He advocates for an 'imperfect shipping' cycle (80% functionality) to align with the rapid evolution of reasoning models, transitioning the user experience from keyword search to natural question-based workflows.
Perplexity CEO Aravind Srinivas frames their Comet browser not as a Chrome competitor but as a new product category — an "agentic browser" that functions like a mini-OS by managing background processes, state, memory, and cross-app orchestration in natural language. The browser's architectural advantage over chat-based agents (e.g., ChatGPT Operator) is that it never requires credential handoff — it operates within the user's existing authenticated session. Srinivas positions current agentic capability as analogous to Perplexity's pre-GPT-4 answer engine quality: functional but rough, with the bet that model improvements will compound on top of the already-built infrastructure. The business model thesis is that browser-native commerce and a tiered subscription stack (up to $2,000/month) can generate tens of billions in revenue without advertising.
Aravind Srinivas argues that the "chat layer" race is already won by OpenAI, and the only viable path to market leadership is owning the browser — the actual entry point for ~15 billion daily queries. Comet, Perplexity's Chromium-based browser, is designed around a hybrid client-server architecture: user authentication and third-party session data stay local, while frontier model inference runs server-side, directly addressing the security liabilities of cloud-hosted agent approaches like OpenAI's Operator. The strategic thesis is that agentic workflows — autonomous browsing, task completion, and orchestration across tabs — can only be delivered natively from within the client, not from a chat window or hosted environment. Srinivas sees Google as structurally constrained from shipping aggressive AI agents at scale due to the direct conflict with its Adwords revenue model and the reputational risk of errors in Chrome.
Perplexity CEO Aravind Srinivas positions Comet not as a browser replacement but as a client-side agentic runtime — executing multi-step workflows (price comparisons, email triage, research compilation) without exposing user data to Perplexity's servers. The architectural bet is that keeping data on the client while invoking frontier models on-demand is both a privacy and a competitive differentiator versus MCP-based desktop agents. Long-term vision is a "task manager" layer of persistent background processes that effectively functions as an AI operating system. With 500K+ waitlist signups and OpenAI rumored to be building a competing browser, Perplexity is racing to lock in retention before scaling monetization via usage-based pricing rather than advertising.
Aravind Srinivas argues that AI has plateaued as a tool layer and must evolve into a workflow/process automation layer — and that the browser is the necessary (not merely sufficient) runtime for this transition. Comet, Perplexity's forthcoming browser, is framed as a "cognitive operating system" that embeds AI into the omnibar, sidebar, and new-tab interface, enabling autonomous task execution within existing authenticated user sessions. The core thesis is that persistent memory accumulated through browser activity is the only credible path to proactive, personalized AI — surpassing what chat apps or OS-level integrations can capture. Srinivas also positions Perplexity as a structural independent counterweight to Big Tech consolidation in AI, with Bloomberg Terminal disruption and financial research as a near-term monetization vector.
Perplexity CEO Aravind Srinivas positions Comet not as a browser but as a personal AI agent that operates asynchronously — drafting emails, scheduling meetings, and prioritizing tasks without explicit user prompting. The core architectural insight is that future AI value delivery requires end-to-end orchestration of multiple specialized models (reasoning, context retrieval, synthesis, computer use) rather than a single generalist model. Critically, Comet addresses the deep personalization problem — knowing who matters in your life — by pulling browser history locally on-device rather than centralizing user data on servers, sidestepping the storage and privacy concerns that otherwise give incumbents like Google a structural advantage.
Perplexity AI, valued at $9 billion, is challenging traditional search engines by prioritizing accurate, cited answers over virality. CEO Aravind Srinivas emphasizes eternal curiosity and continuous learning as crucial in the age of AI. The company focuses on building trust through verifiable information and aims to integrate AI into browsing for a more active, agentic user experience.
Aravind Srinivas argues that Google's structural inability to cannibalize its own ad revenue is the core reason it cannot fully ship an AI-native search product, despite having superior infrastructure. Perplexity is exploiting this gap by positioning itself as an "accuracy layer" for AI decision-making and expanding distribution through hardware OEM partnerships and a forthcoming AI-native browser ("Comet"). The browser strategy is framed not as a Chrome clone but as a "cognitive operating system" that blends navigation, information retrieval, and agentic action in a single interface. Revenue sharing with publishers, rather than pure traffic referrals, is pitched as the sustainable content economy model for the AI search era.
Perplexity CEO Aravind Srinivas argues that the decisive AI battleground is not search itself but the browser's omnibox — whoever controls that interface will capture all traffic, just as browser-integrated search consolidated the search market two decades ago. Google, despite deep AI expertise and foreknowledge of disruption, is structurally trapped: its search advertising business runs at 80–90% margins and funds everything else, making it nearly impossible to cannibalize with a lower-margin AI product without existential financial consequences. Srinivas contends that true moat in AI will come not from raw model capability (which commoditizes) but from deeply personalized, context-rich AI assistants that users invest time training — making switching costs high and willingness-to-pay well above current subscription pricing. Execution quality, obsession with detail, and the absence of legacy constraints remain the primary advantages available to lean challengers.