Perplexity Deep Research

Perplexity Deep Research is what happens when you let an LLM plan its own search strategy instead of forcing you to prompt twenty times. You pose a question, it breaks that question into sub-queries, runs parallel searches across its index, then writes a report with inline citations. The autonomy here is real: I've handed it "What's changed in European GDPR enforcement since 2022?" and come back ten minutes later to a 2,000-word brief with twenty-odd sources, organised by theme. That's the workflow it replaces: open fifteen tabs, skim, copy quotes, reconcile contradictions, cite. The free tier gives you five deep research queries a day, which is enough for most people who aren't professional analysts. Where it shines: broad scans where you don't yet know the sub-questions. I used it to map the state of WebAssembly tooling before a client pitch. It surfaced compiler chains I hadn't heard of and grouped them sensibly. The citations are genuine links, not hallucinated, and it flags when sources conflict. That's unusual. Most LLMs either invent references or bury disagreement. Failure modes: it's only as current as Perplexity's index, which lags by hours or days depending on the topic. If you need this morning's news, standard Perplexity search is faster. It also can't access paywalled journals or proprietary databases, so academic literature reviews hit a ceiling quickly. The writing is competent but flat; you'll rewrite the intro and tighten the argument before you ship it to a human audience. And because it's autonomous, you can't steer mid-flight. If it misunderstands your question, you start over. Versus competitors: ChatGPT's search is faster but doesn't decompose questions or synthesise at this depth. Claude with search tools requires you to prompt the decomposition yourself. Consensus and Elicit are better for pure academic lit review but narrower in scope. Perplexity sits in the middle: broad enough for business research, structured enough to feel like an analyst did the legwork. The free tier is generous enough that most users won't need to pay. If you're running ten deep dives a day, the Pro subscription makes sense, but at that volume you're probably already a customer.