Perplexity AI Review

Roughly $200M ARR against a $20 billion valuation. The math sits at 100x revenue — that's the price you pay to believe Google's AI Overviews and ChatGPT search won't close the gap in 24 months.

Comet Enterprise launched March 2026 with CrowdStrike Falcon integration, MDM deployment, and a no-training-on-enterprise-data commitment. Fortune, AWS, and Bessemer signed early. That's the real product to evaluate now — the $20 Pro plan is the consumer wedge.

But the catch is the multiple. Nvidia, Bezos, and SoftBank backing buys runway, not inevitability, and Anthropic's Claude with web search is the obvious fallback if positioning shifts. Pilot Comet Enterprise on one team for 60 days. Don't standardize the Pro seats across the org until the browser story actually holds with your security review.

I've been using Perplexity daily for about 14 months now, initially as a personal tool but increasingly for technical research. What hooked me was the citation system - when researching new technologies or vendor claims, I can quickly verify sources without diving through multiple tabs. It's particularly strong for API documentation searches and understanding emerging tech trends.

The lack of enterprise controls is my main frustration. We can't centrally manage accounts, there's no SSO, and I can't enforce data retention policies. The API is still in beta and rate limits are restrictive for any serious automation. That said, the Pro version has saved me hours weekly. The ability to search across academic papers and get synthesized answers with proper citations has been invaluable for evaluating new architectural patterns.

The category split is real. Google and ChatGPT search layer answers onto a chat surface; Perplexity made the answer engine the product and shipped Comet as the browser around it in October 2025. For a Head of Knowledge, that's not a feature — it's a bet that research lives in a dedicated client.

Sonar API runs $3 per 1M input tokens and $15 per 1M output. Enterprise Pro at $40 per seat adds SSO and 500 daily research queries — real procurement shape. Spaces and the Academic mode signal editorial intent: citation-first, source-graded, closer to a research librarian than a search box.

But the catch is the substrate. Perplexity routes to GPT-4o, Claude, and its own Sonar models — the moat lives in orchestration and crawl, not the weights. If OpenAI ships a comparable cited-answer surface natively, the differentiator narrows fast. The 3-year bet works as a research client, not a search replacement.

I've been using Perplexity daily for over a year now, primarily for researching technical issues and finding up-to-date documentation. What sets it apart is how it combines search with AI responses while actually citing sources - something I desperately need when evaluating new libraries or debugging obscure errors.

The web interface has become part of my workflow, especially when I need recent information about framework updates or best practices. The Pro subscription's ability to switch between models (GPT-4, Claude) is surprisingly useful depending on the task.

My main frustration is the limited API access. While the web product is stellar, integrating Perplexity's capabilities into our internal tools would be game-changing, but the API is still restricted to select partners.

I started using Perplexity AI for quick competitor research, but it's evolved into something much more valuable for our marketing operations. Every morning, I use it to scan industry news, validate content ideas, and dig into customer pain points we're addressing. My content team loves how it surfaces sources – we've cut our research time by about 40%.

What surprised me most is how it's improved our thought leadership. Instead of generic blog posts, we're creating deeply researched pieces backed by real data. The Pro version's ability to search academic papers has been gold for our whitepapers. Yes, it's not built for marketers specifically – there's no campaign tracking or lead scoring. But as a research and ideation engine? It's become irreplaceable.

I've been using Perplexity AI daily for over a year now, primarily for market research, competitor analysis, and quick fact-checking before board presentations. What sold me initially was the citation feature - being able to verify sources instantly has saved countless hours of manual validation. The $20/month Pro plan fits comfortably in our tools budget, and I've gotten three team members on it.

The real value comes from speed. Yesterday I needed comparative pricing data across five industries for a strategic planning session - what would've taken an analyst half a day took me 20 minutes. However, measuring concrete ROI is tricky since it's mostly about time efficiency rather than direct cost savings. My main frustration is the lack of enterprise invoicing - we still can't get consolidated billing for our team accounts, which makes expense tracking messier than it should be.

Focus modes are where a working researcher lives. Switching a query to Academic pulls Semantic Scholar and arXiv results into the synthesis instead of SEO blogspam — friction every Google-to-scholar-tab workflow knows. Sonar handles the routing without a rephrase.

Spaces hold persistent context across queries, with memory that builds the more you work in one. Pro lands at $20/month with 20 Deep Research queries per day and $5 in Sonar API credits — roughly 200 Pro queries. However, follow-up threads drift past 15 turns on dense topics, where Claude's Projects keeps the thread tighter.

Docs read like the team uses the API daily — Sonar at $1/$1 base versus $3/$15 Pro, with request-fee math right there. The catch is citation depth. Perplexity surfaces 5-8 sources per answer, but the synthesis sometimes glosses a paper's actual claim — a daily fight for anyone doing literature review.

I started using Perplexity for work research and now I'm hooked. What sold me was getting actual answers with sources instead of wading through SEO-optimized articles. I use it dozens of times daily - checking facts for presentations, understanding technical concepts, even settling dinner debates.

The interface is refreshingly simple. Type your question, get an answer with numbered sources. The Pro version's ability to search academic papers has saved me hours. My only gripe is the mobile app feels limited compared to desktop - no collections, harder to manage threads.

It's not perfect for everything. Sometimes I need to rephrase questions to get useful answers, and it occasionally misses nuance in complex topics. But for daily research tasks, it's become indispensable.

I was an early adopter of Perplexity Pro, drawn by the promise of accurate, sourced answers. For months, it was my go-to for research and fact-checking. But somewhere around the 8-month mark, things went downhill fast. They keep redesigning the interface every few weeks, breaking muscle memory and hiding features I relied on. The Collections feature, which I used to organize research projects, now randomly loses documents. Half the time, the AI gives me outdated information despite claiming to search the web. The final straw was when they removed the ability to export conversation history properly – I lost months of research notes. Their support team just sends canned responses about 'exciting updates coming soon.' I've switched to Claude with web search plugins and haven't looked back.