Perplexity

What it is

Perplexity is an AI answer engine: ask a question, get a synthesized answer with inline citations you can click through to the source. Around that core sit Spaces (persistent research projects with pinned sources), the Comet browser with an on-page assistant, the Sonar API for developers, and Enterprise tiers with SSO and internal-data connectors. It’s the default tool for ops folks who want grounded, cited answers — not raw model output — when researching companies, regulations, or candidates.

Why it shows up in ops stacks

• Grounded by default. Every answer lists its sources, which matters for ICP research, competitor intel, and any output headed to a CRO or GC who will ask “where did this come from?”
• Spaces for project research. Pin sources, re-run queries against them, and share the Space with the team — better suited to an ongoing research workstream than a one-off ChatGPT thread.
• Real-time web, built in. Claude and ChatGPT both search now, but Perplexity’s whole UX is organized around retrieval and citation rather than bolting it onto a chat.
• Official MCP server. Perplexity ships a first-party MCP server (@perplexity-ai/mcp-server), so you can wire Sonar search and deep research into Claude, Cursor, or your own agent instead of copy-pasting answers out of the app.

Pricing

• Free — roughly 5 Pro searches/day plus unlimited basic search. Fine for occasional lookups.
• Pro — $20/mo ($200/yr): unlimited Pro Search, ~20 deep-research runs/day, access to the frontier model picker, and the Comet assistant. This is the tier most individual operators want.
• Max — $200/mo ($2,000/yr): removes research/Labs limits and adds Model Council (one query fanned across multiple frontier models with a synthesized answer) plus the highest file and storage caps. A power-user tier, not a team default.
• Enterprise Pro — $40/seat/mo ($400/seat/yr): SSO, SCIM, admin controls, SOC 2 Type II, Internal Knowledge Search over your own docs, and — critically — your data is not used for training.
• Enterprise Max — $325/seat/mo ($3,250/seat/yr): Enterprise Pro plus the highest limits and model access.
• Sonar API — usage-based: $1 per million tokens at the base tier, up to $3 input / $15 output per million for Sonar Pro, plus a per-request web-retrieval fee ($5-12 per 1,000 requests).
• The Comet browser is free; Comet Plus (~$5/mo, bundled into Pro and Max) adds premium publisher content.

Best for

• RevOps building account-research dossiers with sources a rep can defend on a call
• Recruiters validating market and comp data before it goes into a hiring plan
• Legal ops scanning regulatory updates and case summaries — with the click-through-to-source discipline below

Alternatives — and when to pick them instead

• ChatGPT (with search) — the market-share leader. Pick it if your team already lives in ChatGPT and wants one tool for both retrieval and long-form drafting rather than a dedicated answer engine.
• Google Gemini / AI Mode — the other giant, and the fastest-growing entrant in consumer answer-search. Pick it if you’re standardized on Google Workspace and want AI search inside the same account and data boundary.
• Claude — pick it when the job is reasoning, synthesis, or drafting over sources you supply, not open-web retrieval. Different job; pair the two rather than cross-shopping on price.

Watch-outs

• Citations look authoritative but can misattribute. The model sometimes cites a source that doesn’t actually support the sentence. Guard: click through every load-bearing citation before it lands in a board deck, GC memo, or candidate brief.
• Retrieval engine, not a writing engine. For long-form reasoning and polished drafting, Claude/ChatGPT are stronger. Guard: use Perplexity to gather and cite, then hand the raw material to a writing model for the deliverable.
• The $20 → $200 jump is steep. Max’s Model Council is real, but most ops users never hit Pro’s research limits. Guard: stay on Pro until you’re routinely capping out deep-research runs before paying 10x.
• Only Enterprise tiers wall off your data. Free and Pro don’t come with the no-training guarantee, SSO, or admin controls. Guard: never put company or candidate data into Free/Pro; require Enterprise Pro before any team rollout.