Tagged
Perplexity News
Releases, search features and business moves from Perplexity. Curated and summarized from dozens of sources by AIBriefs.
Perplexity built CobbleDB with AI agents to replace DynamoDB
Two engineers and hundreds of persistent coding agents built CobbleDB, a key-value database serving Perplexity search, in two months. Perplexity says migrating off AWS DynamoDB will save up to $100 million yearly; the agents wrote code but were not allowed to run the database.
Perplexity ships pplx-search-sdk cookbook for coding agents
Perplexity Computer launches pre-installed on HP ZBook Ultra G3a
Crusoe signs multiyear cloud deal with Perplexity AI
Data center operator Crusoe will rent AI chips and related software to Perplexity AI under a multiyear cloud capacity agreement. The deal continues Crusoe's push to win more business renting out AI infrastructure.
Perplexity Portable Computer launches on Windows RTX PCs
Perplexity's local agent runs on GeForce RTX PCs and RTX PRO Workstations, building on existing DGX Spark and Linux RTX support. It ships with a local model such as Qwen 3.8 27B post-trained for Perplexity Computer, and locally completed work doesn't consume Perplexity Computer credits.
Perplexity CEO: Computer becoming a workspace of humans and AIs
Perplexity Computer makes you money
Perplexity API MCP server adds OAuth sign-in
Perplexity introduces Q2D-Web benchmark for agentic RAG retrieval
Perplexity Search API now available in Hermes Agent
Perplexity moves inference to NVLink Blackwells
Perplexity details GPU embedding stack: Ivy, Tulip, ROSE
Perplexity published research on its serving infrastructure for pplx-embed, combining Ivy, Tulip, and ROSE to lower latency and improve throughput across online and batch workloads. The stack reduces cost compared to off-the-shelf solutions.
Perplexity CEO explains flipped Q&A model at HubSpot Live
Aravind Srinivas, CEO and co-founder of Perplexity, discussed why the company inverted the traditional question-and-answer model, arguing that trustworthy answers enable better follow-up questions. He spoke at HubSpot's UNBOUND conference.
Perplexity CEO discusses local vs. cloud AI computing
Aravind Srinivas, Perplexity CEO, told CNBC that people want to use their own local hardware for AI, discussing the trade-offs between local and cloud computing and data centers.
Perplexity open-sources Numbat agent forensics tool
Perplexity CEO Aravind Srinivas tweets hiring call for inference and infrastructure…
GPT-6 Astra comes to Perplexity Computer for Pro and Max
Perplexity details Ivy HTTP gateway for inference
OpenAI's GPT-6 Astra praised by Perplexity CEO as frontier model
Study: Perplexity cites low-quality, machine-generated sources
Across 380 software categories, 59.8% of Perplexity's grounded citations point to domains ranked below #100,000, with 23.4% not in the top million. Three sites published 215,128 machine-generated "best software" pages, none existing before December 2023.
Perplexity Agent API adds 42 models including GLM 5.3
Perplexity CEO teases 'Private, Personal, Powerful AI'
Perplexity Computer adds Coinbase, OpenSea, and Public connectors for agentic trading
Perplexity adds Fable as orchestrator on Perplexity Computer
Perplexity launches hybrid compute for Mac app
Perplexity Computer on Mac can now split tasks between cloud models and local models, keeping sensitive data on-device. Available today in the Mac app. Powered by PPLX Qwen 3.8 27B, with a Privacy Gate to detect PII before sending to cloud.
Perplexity CEO claims best search at any compute level
Decagon uses Perplexity search for Delta, Ticketmaster support
Perplexity Computer launches with Brain memory and Dream agent
LangChain builds auditable VC research agent with Perplexity
The agent drafts a cited investment memo in ~90 seconds for ~$0.40, using Perplexity Agent API, LangGraph, and LangSmith evals. It runs four parallel research nodes (team, financials, product, market) then a synthesizer writes a seven-section memo with citations.
Perplexity to release local-first model benchmarks
Perplexity unveils PPLX 27B, trained inside Computer harness
Perplexity launches Portable Computer for NVIDIA DGX Spark
Portable Computer runs the entire agent runtime locally on DGX Spark, with zero per-token cost for local steps. The on-device 27B model scores 82.6% on real knowledge work, beating open-source harnesses Pi and Hermes; post-trained PPLX 27B reaches 85.4%.
Andrew Gordon Wilson joins Perplexity research team
Cheshire Academy uses traffic-light system for classroom AI rules
The Connecticut private school, about 400 students in grades 9-12, lets instructors choose whether to use AI and trains staff on general prompting techniques rather than mandating specific tools. Teachers use ChatGPT, Perplexity and educator platform MagicSchool; librarian George Aiello says the "vast majority" of instructors use AI in some way.