Perplexity
Built for real-time research with citations. The go-to when you need to answer a current-market or regulatory question AND ship the source links with the answer.
Best for
- Synthesizing this week's market commentary across multiple outlets — pulls from current news with inline citations you can click and verify
- Regulatory and rule-tracking questions ('what's the current FINRA position on AI-generated content disclosure?') with sourced answers in under a minute
- Pre-meeting research on a prospect firm, recent news, and leadership changes — Spaces let you organize the research per prospect for repeat reference
- Fast fact-verification — faster than a Google rabbit hole because the answer comes with sources, not ten blue links to evaluate
Avoid
- Drafting client communications — Perplexity is a research tool, not a writer; output reads like a Wikipedia summary, not an advisor's voice
- Analyzing your own client data or running calculations — no Code Interpreter equivalent; use ChatGPT or Claude for that
- Anything that doesn't need fresh sources — for evergreen knowledge or pure reasoning, Claude or ChatGPT are usually faster and often deeper
Shortcuts
- Focus modes (All, Academic, Writing, YouTube, Reddit, Math) constrain the source pool — Academic for whitepapers, Writing for help-me-draft tasks, Math for calculation-backed answers
- Spaces save a topic-specific context (e.g., 'tax law 2026', 'a particular client's industry') so follow-up questions re-use prior research. Personal Spaces work on Pro; Shared Spaces (team collaboration, admin controls) require Enterprise Pro or higher.
- On Pro, you can choose which underlying LLM powers Perplexity reasoning (Sonar default, Claude, or GPT-class models) — Claude tends to win for compliance-sensitive synthesis
- Threads are conversational follow-ups — ask 'expand on the second point' rather than rephrasing the whole original question
Model variants
Perplexity wraps multiple underlying LLMs. Default mode uses Perplexity's own Sonar model (fast, citation-tuned). Pro mode lets you select Claude, GPT-4 class, or specialized models per query. The differentiator is the retrieval-and-citation layer built on top, not the model.
Compliance gotchas
- Free-tier queries may be used for product improvement — never paste client PII or specific account details. Use generic phrasing ('a 65-year-old retiree with a $2M portfolio') for any client-specific scenario.
- Perplexity Enterprise Pro ($40/user/mo and up) offers zero data retention and SOC 2 — required for any client-related research. Consumer Pro ($20/mo) is fine for your own learning but not firm data.
- Citations are auto-generated and occasionally misattributed (a paraphrased point pointing at the wrong source). Always click through before quoting a source to a client or in writing.
- Perplexity doesn't produce 'compliant' output — it's a research tool. Any answer that ships to a client still needs your firm's normal compliance review, especially around forward-looking statements pulled from market commentary.
Compliance posture last verified 2026-05-29. Full posture matrix →
Pricing
Free tier (limited Pro searches per day, no model picker). Pro $20/mo (more searches, model picker, file upload — INDIVIDUAL only; no shared Spaces or admin controls). Max $200/mo (heaviest individual use; 4,000 Pro searches/week + 500 Research queries/mo). Enterprise Pro $40/user/mo ($400/year per seat) — adds zero retention, SOC 2, Shared Spaces, SSO/SCIM, Internal Knowledge Search (up to 500 shared files). Enterprise Max $325/seat/mo for highest-volume teams + expanded audit logs + custom retention. Standalone Sonar API priced per-token for embedding research in your own tools. (SCIM + audit logs require 50+ members or 1 Enterprise Max user.)