The Perplexity Blog

Two stacks of four browser-page views represent an attacked page and its clean twin, with the injection present only in the attacked stack.
ResearchNov 2, 2026BrowseSafe++: Building stronger defenses against real-world prompt injectionAn open benchmark and two lightweight detectors for browser agents.
Woman seated on a sofa using a smartphone.
NewsNov 3, 2026Computer for e-commerce is built for brand growthUse Computer for e-commerce to analyze Shopify sales, build product pages, and create ad and email campaigns for growth.

Featured

  1. Man pulling back curtain on horizon
    IdeasNov 1, 2026The Knowledge Talk
  2. Abstract glass panel with illuminated edges separating light and shadow against a dark background.
    NewsOct 15, 2026Open-sourcing SPACE, the sandbox platform behind ComputerPerplexity open-sources SPACE, the sandbox platform behind Computer, to help developers build safer autonomous AI agents
  3. Two people working at separate desks in a dimly lit office.
    DevelopersSep 28, 2026Agent API now supports reusable agentsTeams create reusable agents with versioned Profiles, Skills, and managed connectors they can share across applications.

More from the Perplexity Research team

  1. A good teacher meets the student where they areA new self-distillation method that jointly trains teacher and student to improve model performance and training efficiency on complex reasoning and agentic tasks.
  2. Building shared workspaces for people and agentsHow Perplexity Projects keep long-running work persistent, concurrent, and traceable.
  3. Multimodal embeddings beyond a single vectorA new family of late-interaction models for text and image retrieval, with a shared embedding space across model sizes.
All research

Recently published

217 posts found

  1. Autonomous agents: what they are and how they workThis guide explains what autonomous agents are, how they plan, use tools, remember, and check their work, and how people use them for research, email, coursework, and shopping.
  2. Computer for e-commerce is built for brand growthUse Computer for e-commerce to analyze Shopify sales, build product pages, and create ad and email campaigns for growth.
  3. BrowseSafe++: Building stronger defenses against real-world prompt injectionAn open benchmark and two lightweight detectors for browser agents.
  4. AI agent memory: How to build AI agents with persistent contextThis guide explains what AI agent memory is, how working, episodic, semantic, and procedural memory work together, and how to build and test agents that improve through persistent context.
  5. The Knowledge Talk
  6. How to Fine-Tune an LLM [and When to Use RAG or Prompting Instead]This guide explains how to fine-tune an LLM step by step, compares full fine-tuning with LoRA and QLoRA, and shows when prompting or RAG is the better way to improve a model's outputs.
  7. Text embeddings explained: how they work [+examples]This guide explains what text embeddings are, how they turn text into vectors that capture meaning, and how to choose, evaluate, and apply embedding models to search, RAG, and other real-world use cases.
  8. New FedRAMP certification brings Perplexity to more government workflowsFederal agencies can search the live web with Perplexity and get cited answers as part of a new FedRAMP certification.
  9. Basic RAG Implementation: A Step-by-Step TutorialThis beginner-friendly tutorial explains how a basic RAG implementation works, covers best practices, and walks through building a working command-line RAG system with Python, ChromaDB, and Perplexity's embedding and language models.
  10. How does an LLM gateway work? A practical guide for engineering teamsAn LLM gateway is a reverse proxy that routes, authenticates, caches, and logs your model calls; this guide follows a request through one, covers the security and reliability risks to plan for, and compares building, self-hosting, managed, and API-gateway options.