The research-to-deck stack

All three tools here have a genuine free tier, so a working configuration costs nothing while you evaluate it. The 03 COSTS table above breaks down each tool's pricing. Three meters climb with use: Perplexity's deep research runs, NotebookLM's daily generation caps, and Gamma's deck volume. Watch Gamma's credits especially; its Agent and premium models burn them beyond the sticker price, so the published figure is not the whole bill. Treat each upgrade as an independent decision, triggered by a different kind of work.

No. Perplexity alone answers most research questions with citations attached, and many jobs stop honestly there. The numbered steps run in pipeline order, which doubles as the adoption order: start with Perplexity, add NotebookLM once your sources grow into a corpus worth interrogating rather than a handful of links, and bring in Gamma at the point conclusions need presenting to someone other than yourself.

Your Perplexity subscription already owns discovery: current, sourced answers and Deep Research reports from the live web, so keep leaning on it there. What it does not do is master a corpus you have assembled: that is NotebookLM, answering strictly from your uploaded material with citations that jump to the passage. Gamma then carries the final stage, turning your conclusions into a presentable deck.

Perplexity and NotebookLM both answer questions with citations, the one real overlap here. The dividing rule is where the answer comes from. Perplexity wins on the open web: live sources it finds for you while you are still discovering. NotebookLM wins on the closed corpus: it answers only from the material you have uploaded, with citations that trace to the exact passage. The Perplexity and NotebookLM comparison, linked above, covers the finer calls.

Two signals, both about scope rather than price. The first is when your conclusions have to withstand real scrutiny, and gathering plus a personal notebook no longer gives you a defensible, documented evidence trail. The second is when several people work the same body of sources and need shared, verifiable ground instead of individual notebooks. Both point to the competitive tier, the evidence-grade research stack.

Treat the free and consumer tiers as semi-public: Gamma's training terms are weak or unverified, NotebookLM's uploads sit on Google servers, and Perplexity guarantees no training only on Enterprise. Keep client strategy, confidential records and credentials out until each tool earns them: Perplexity's Enterprise Pro/Max tier for an unconditional no-training guarantee (or the AI data retention toggle switched off on Pro), NotebookLM under checked institutional or Workspace terms, Gamma on a paid plan with a DPA. Human-review anything generated before it leaves your hands.