Tom Osman
[ RETURN_TO_INVENTORY ]

Perplexity

Answer engine + API

Perplexity homepage
Live preview captured from perplexity.ai

Perplexity is my research copilot when I want sourced answers fast. It runs queries across the live web, applies its LLM reasoning stack (currently powered by models like pplx-70b, pplx-7b, GPT-4o, Claude, etc.), and returns concise responses with inline citations and follow-up pivots. It also offers Pro features (File Uploads, Focus, Pro Agent) and a developer API. (perplexity.ai)

Why it’s in my toolkit

  • Trustable citations. Every answer links to the underlying pages, so you can drill into the original sources before publishing or acting on the info.
  • Follow-up pivots. Ask iterative questions, pivot to specific sources, or switch “Focus” modes (Academic, Writing, YouTube, Reddit, Code) when you need a particular corpus.
  • Pro Agent. Perplexity’s agent can run multi-step queries, upload files/PDFs for cross-reference, and render sharable “Collections” of research.
  • API + pplx models. Perplexity exposes pplx-hosted models + search endpoints, letting you embed the same answer engine into your apps or agents.

How I use it

  1. Video research. Pull the latest stats, methodologies, or product analyses with citations before recording.
  2. Agent seeding. Grab a curated list of sources to prime my own agents or summarizers.
  3. Shareable briefs. Package investigations into Perplexity “Collections” so collaborators can dig in without re-running queries.

Getting started

  • Use the web/mobile app at perplexity.ai for free queries; upgrade to Pro for GPT-4o/Claude access, Pro Agent, and higher limits.
  • Enable “Focus” modes for niche searches (Academic, Writing, Code, etc.).
  • Explore the developer docs for pplx APIs if you want to integrate Perplexity search or hosted models into your stack.
  • When citing Perplexity outputs publicly, double-check the linked sources—they make it easy to verify, but accuracy is still on you.
Open Perplexity