How Perplexity can Improve Research Productivity

Most people use Perplexity like Google. That’s a mistake. Used correctly, it replaces hours of analyst work. Here are 7 prompts that turn Perplexity into your personal research team. 1. Market timing intel   Prompt:   “Find every major announcement, funding round, and product launch in [industry] from the last 90 days. For each one, show the date, companies involved, dollar amounts, and what larger trend this signals.” Use this when you want to understand where capital and attention are actually moving, not where headlines point. 2. Competitive teardown   Prompt:   “Deep dive on [company name]. I need their real revenue model, customer acquisition strategy based on hiring and ads, product positioning, and where they are most vulnerable.” This is how you reverse-engineer competitors without guesswork. 3. Trend validation   Prompt:   “I’m seeing buzz about [trend]. Separate hype from reality. Show how many companies are building vs talking, how much money is being invested with dates, and what technical blockers still exist.” Great for founders, operators, and investors who want signal, not noise. 4. Customer research goldmine   Prompt:   “Find real conversations about [problem] across Reddit, X, forums, and review sites from the last 12 months. I want direct quotes. Group them by recurring pain points and language patterns.” This replaces surveys and gives you copy straight from the market. 5. Policy and regulation radar   Prompt:   “Track regulatory changes affecting [industry] globally. Show proposed legislation, recent court cases, timelines, likelihood of passing, and which countries are moving fastest.” Critical if regulation can make or break your roadmap. 6. Talent flow mapping   Prompt:   “Show where senior people are moving in [industry]. Who left major companies, where they went, which startups are hiring aggressively, and for which roles.” Talent moves before markets do. 7. Price intelligence   Prompt:   “Analyze pricing across [market]. For the top players, show exact tiers, feature differences, pricing changes over the last two years, and what patterns reveal about positioning.” This is how you price with context, not intuition. If you’re using AI only to write faster, you’re missing the leverage. If you’re using it to think better, you’re ahead. Follow @Amit Kumar Soni for more curated AI for Good Knowledge

Been running Perplexity Pro, ChatGPT Plus, Claude and Google’s Gemini + NotebookLM for core research and analysis, comp intel, internal synthesis and decision support. Some clear patterns now. Perplexity Pro: best for OSINT-style research. Used it to track product pipelines, market research, regulatory filings, regional pricing updates. Fast, well-cited, low friction. Weak on cross-source synthesis but great for narrowing search space fast. Claude Opus: strongest for long-context distillation. Took 80+ pages of BU plans and call notes, mapped out structured insights with minimal drift. Used it for parsing customer transcripts, internal decks, layered feedback docs. High context integrity, slower output, low hallucination risk. ChatGPT Plus: most adaptable. Used it for rewriting POV docs, turning growth dashboards into briefings, processing CSV dumps, cleaning up call logs. Vision input helps extract from screenshots, charts, PDF tables. Code interpreter useful for sanity-checking metrics and running quick EDA on Excel. Google Gemini 1.5 + NotebookLM: light for context recall, improving inside Workspace. Gemini helps with GDocs summarisation, note clean-up, Drive consistency checks. NotebookLM better for multi-doc prep, layered Q&A, and thematically slicing dense material. Still not ideal for external research but solid for internal doc workflows. Current working stack: - Perplexity for live intel and signal surfacing - Claude for input-heavy distillation and clarity - ChatGPT for structured synthesis, writing and data analysis - Google (Gemini + LM) for doc-bound prep and Workspace ops LLMs are more than intelligent assistants, they’re context filters, reasoning engines and time multipliers when matched well to the task. If you’re using a different config across product, strategy or ops, would be great to exchange notes. #GenerativeAI #LLMWorkflow #ProductivityTools #AppliedAI

If you are in a junior PR role, this is a defining moment. AI is going to impact your role. It will especially impact (negatively) those that operate without clear ownership, defined systems, or measurable contribution. Who is most exposed? • Media list builders • Writers who don't impact or own strategy • Those monitoring reactive sites: HARO or Qwoted • Support roles not tied metrics or impact • Reporting assistants (e.g. clip books) • Entry-level support who “help” but do not own a workflow There is a next-level move. Take AI by the horns and use it to your advantage. Become a workflow master who can build the infrastructure layer and execute it -- and you have immediate value and impact. Here are some ideas to get you started: 1: Redefine What “Media List” Means Build dynamic intelligence. Identify: • Active writers in the last 7 days • Journalists increasing frequency on a topic • Reporters shifting narrative angles • Recent outlet moves • Competitor citations Set up a recurring research prompt inside Perplexity AI: “Identify journalists who have published in the past 7 days about [topic]. Summarize their angle, headline framing, sources cited, and whether competitors were mentioned. Highlight any new bylines or outlet changes.” 2: Build a Visibility Intelligence Workflow Replace the clip book with a structured visibility system. Create: • A coverage tagging system (narrative theme, product, executive, tier) • A key message penetration tracker (Y/N per article) • An executive quote density score • A competitor co-mention tracker • An outlet influence score (AI-cited vs non-AI-cited sources) Automate: • Drop links into AI for tagging and summarization • Generate monthly narrative share charts • Flag weak message penetration automatically 3: Redefine Reactive Monitoring Optimize performance. Build a Reactive Performance Optimization Workflow. Create: • Centralized opportunity log • Topic tagging system • Spokesperson win-rate tracker • Outlet ROI categorization (high, medium, low yield) • Time-to-response tracking Automate: • AI-generated first-pass drafts • Monthly conversion summaries • Auto-identification of high-performing topics Output: A monthly Reactive Conversion Report that shows which topics and outlets deserve continued investment. 4: Redefine Thought Leadership Build a positioning engine. Create: • Quarterly theme calendar aligned to business priorities • Industry narrative tracker (dominant themes in coverage) • Content gap log (topics competitors are not addressing) • Repurposing workflow (article → pitch angle → social → newsletter) Automate: • AI clustering of industry articles • Executive transcript summarization • Headline framing comparison CREATE YOUR PLACE BY THINKING LIKE THIS: Static → Dynamic Task → System Activity → Measurable impact AI will not replace the person who builds and owns the workflow that drives outcomes. #advice #business #PR #communications #junior

I run a £20K/month research process through Perplexity for free. Founders don't realise what they're sitting on: When I was building Searchable, I needed: - Deep market analysis on AI search adoption - Competitive intelligence on legacy SEO platforms - Regulatory landscape research on AI governance Hiring analysts would've cost £15K-£25K per project. Instead, I built a prompting system that turned Perplexity into a Harvard research team. Here's the system I use to get the most out of Perplexity: ✅ Time-bound everything → "Last 24 months only" filters recency bias ✅ Demand structure → Tables, headings, sections beat wall-of-text responses ✅ Force citations → "Cite all claims" makes output defensible ✅ Require both sides → Evidence for, against, and what's uncertain ✅ End with action → "So what?" or "What should someone do next?" ✅ Apply domain knowledge → Review sources, validate, add your expertise And the prompts to get started: (Find the full prompts in the sheet 👇) 1. Market Landscape Snapshot Breaks down market size, growth rate, main players, and 3-5 trends that matter.  Ends with practical "So what?" No speculation allowed. 2. Competitive Comparison Breakdown Creates a positioning table comparing offerings, pricing, and distribution.   Flags contradictions in public information. 3. Trend Validation Check Determines if a trend is real or hype by gathering evidence on both sides.  Verdict: hype, early signal, or established shift. 4. Deep Dive on a Single Question Goes beyond surface summaries.  Shows where experts agree, disagree, and what nuance is missed. 5. Buyer & User Insight Synthesis Analyses real customer language from reviews, forums, and discussions.  Surfaces pain points and desired outcomes. 6. Regulation & Risk Overview Maps the regulatory landscape by region.  Focuses on what affects real decisions, not theoretical compliance theatre. 7. Evidence-Based Opinion Builder Lays out the strongest arguments on both sides.  Highlights where evidence is strong versus mixed. 8. Research-to-Decision Summary Structures output as: what's known, key risks and sensible next steps.  Practical decision framework, not predictions. Market validation that used to take days now takes minutes. And the output quality isn't just faster. It's often better because it's pulling from all sources simultaneously. Have you tried Perplexity for research?  Leave a comment below with your thoughts.  If you want your business to show up in AI tools like ChatGPT and Perplexity,  Searchable is the simplest way to have it done for you. It's an autonomous SEO & AEO Growth Engineer that analyses, fixes, and scales your website to drive customers. Learn more and start your 14-day free trial:  https://lnkd.in/epgXyFmi ♻️ Repost to help your network to level up their research.  And follow Chris Donnelly for more on building with AI.  

I’ll admit it - I’ve always used 𝗣𝗲𝗿𝗽𝗹𝗲𝘅𝗶𝘁𝘆 for one thing: deep research. It’s the LLM I go to when I need precision, sources, and serious brainpower. But it also kind of intimidated me. It felt like the “smarty-pants” of AI - brilliant, but not exactly approachable. Then I read 𝙋𝙚𝙧𝙥𝙡𝙚𝙭𝙞𝙩𝙮 𝙖𝙩 𝙒𝙤𝙧𝙠: 𝘼 𝙂𝙪𝙞𝙙𝙚 𝙩𝙤 𝙂𝙚𝙩𝙩𝙞𝙣𝙜 𝙈𝙤𝙧𝙚 𝘿𝙤𝙣𝙚, and it reframed everything. It’s not just a research assistant. It’s a full productivity ecosystem built around how people think and work. Here are 5 𝘁𝗵𝗶𝗻𝗴𝘀 𝘁𝗵𝗮𝘁 𝘀𝗵𝗶𝗳𝘁𝗲𝗱 𝗳𝗼𝗿 𝗺𝗲: 1️⃣ 𝗔𝗜 𝗮𝘀 𝗙𝗼𝗰𝘂𝘀, 𝗻𝗼𝘁 𝗗𝗶𝘀𝘁𝗿𝗮𝗰𝘁𝗶𝗼𝗻 The guide reframes AI’s first job as protecting your attention - handling the small stuff so you can think deeply again. 2️⃣ 𝗦𝗰𝗮𝗹𝗲 𝗬𝗼𝘂𝗿𝘀𝗲𝗹𝗳, 𝗡𝗼𝘁 𝗬𝗼𝘂𝗿 𝗪𝗼𝗿𝗸𝗹𝗼𝗮𝗱 You don’t need 10 new tools. Perplexity turns your curiosity into leverage - helping you produce at team scale without burning out. 3️⃣ 𝗖𝗼𝗺𝗲𝘁 = 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 + 𝗔𝗰𝘁𝗶𝗼𝗻 I’d only scratched the surface of Comet. Turns out it’s both a “thinking partner” and a “doer” - it can actually execute multi-step tasks for you. 4️⃣ 𝗦𝗽𝗮𝗰𝗲𝘀 𝗳𝗼𝗿 𝗖𝗼𝗻𝘀𝗶𝘀𝘁𝗲𝗻𝗰𝘆 I loved the section on “Spaces” - how you can set brand tone, upload sample work, and ensure everything generated sounds like 𝘺𝘰𝘶. 5️⃣ 𝗦𝘁𝗮𝗿𝘁 𝘄𝗶𝘁𝗵 𝗚𝗼𝗮𝗹𝘀, 𝗡𝗼𝘁 𝗣𝗿𝗼𝗺𝗽𝘁𝘀 Instead of keyword-searching, the guide suggests “thinking out loud.” When you tell Perplexity what you’re 𝘵𝘳𝘺𝘪𝘯𝘨 to achieve, it performs at a strategic level. It reminded me that AI isn’t just a faster researcher - it’s a smarter collaborator. And when you use it to 𝗯𝗹𝗼𝗰𝗸 𝗱𝗶𝘀𝘁𝗿𝗮𝗰𝘁𝗶𝗼𝗻𝘀, 𝘀𝗰𝗮𝗹𝗲 𝘆𝗼𝘂𝗿 𝘁𝗮𝗹𝗲𝗻𝘁, 𝗮𝗻𝗱 𝗴𝗲𝘁 𝗿𝗲𝘀𝘂𝗹𝘁𝘀, it stops being intimidating and starts feeling empowering. Here's the link: https://lnkd.in/girWq5BX 💭 Curious — how do you use Perplexity? ♻️ Repost to help your network. Follow Tonya for AI stuff I use, or try to at least.

You're using Perplexity like Google. That's leaving 90% of its power on the table. Here's what I mean. Most investors open Perplexity, type a question, get an answer, close the tab. That's Google behavior. Perplexity is built for something deeper. A few features most people miss: → Research Mode runs multi-step research. It produces a sourced report. Not a summary. A full IC-style brief. → Spaces let you organize research by theme or ticker. Your semiconductor deep dive stays separate from your healthcare names. → Tasks send you automated briefings. I get daily updates on my watchlist at 7am. I didn't write them. → Internal Knowledge Search lets you upload your own model. Cross-reference it against public sources. One thread. No toggling. → Model Selection gives you 9 different models. Claude for long-form writing. Sonar for citations. Grok for real-time scanning. Most people never switch off the default. This isn't a search engine. It's a research platform. The difference matters when you're covering 30 names and need to move fast. 👉 Save this post. Try it out yourself. P.S. If you want more guides on AI for investors, sign up for my newsletter at felipesinisterra.com

I'm deep into researching a project, need to check if a stat is current, and suddenly I'm 15 tabs deep in a rabbit hole. Sound familiar? I couldn't imagine working without Perplexity Comet at this point. It's not just in the background—it's in the sidecar, right there with me as I work. Comet is the difference between spinning my wheels and actually finishing what I set out to do, with more time for real creative thought. Instead of opening half a dozen sites to verify a source and losing my train of thought, I get answers in seconds without leaving my main task. What's changed everything for me are the shortcuts I've painstakingly built and refined—all living in my Comet sidecar. These aren't just basic commands; they're prompts shaped by years of hands-on experience, newsroom rigor, and my personal standards. The sidecar is where my process lives: every shortcut is an extension of how I approach projects. I keep tuning them, evolving them, because the prompts themselves are my assets—always growing as I do. How I use my Comet shortcuts in the sidecar: - Fact-checking with my editorial standards—assuming nothing, verifying and cross-analyzing everything - Fast background checks and credibility analysis - Brainstorming and riffing in my own style (titles, cold opens, alternate endings, small kernels of an idea) - Real-time context for relevance and timeliness, applied mid-research - Deep quality analysis reflecting my criteria With my Comet shortcuts in the sidecar, my research moves at twice the pace—not because I'm rushing, but because I'm not stopping every few minutes to context-switch. Everything gets checked and organized the way I would do it, just without breaking stride. This is the most engaged I've felt with my work in years. The secret wasn't just better time management—it was building shortcuts that actually match how my brain works. Curious how others use the sidecar and what expert shortcuts make your work genuinely better. https://lnkd.in/g5SFAESv

Apollo.io case study with Perplexity just dropped 🔥 Last year, we found that sales reps using Apollo were spending 5+ hours every week doing research. LinkedIn profile, company site, press releases, back to Google. Repeat. The math doesn't work when you need to research 1,000 accounts at 10 minutes each. So we started building Apollo AI Research (a.k.a. AI Power-ups). But I knew we couldn't build this alone - we needed a partner that wouldn't hallucinate, wouldn't lag, and could scale to millions of queries. Instead of running RFPs or chasing the latest model releases, we did something simpler: we watched what our users were actually doing. Our Director of Engineering, Saravana Kumar, went to a local GTM meetup with 20+ sales leaders. The whole point was to ask one question: what AI tools do you actually use every day? What he found blew me away. More than 50% were already using Perplexity for account research. This was before Perplexity had massive brand recognition. While everyone admitted to using ChatGPT for rewriting emails, half the room had quietly adopted Perplexity for the actual hard work - research. Then a competitor announced their Perplexity partnership for their new GTM platform. That was our signal - time to move. We tested every major AI search solution against the same use cases our reps face daily. Perplexity won by a mile: cited sources you could verify, 2.5X faster than alternatives, and honestly - their team jumping on Slack to fix edge cases in real-time sealed it. We shipped our Perplexity integration and adoption took off: - 46% more meetings booked for users leveraging Apollo AI Research - 0 to 10M enrichments per month in less than a year - 0 to 6K weekly active users who trust AI for their research workflow What's most interesting? That meetup insight was predictive. Turns out AI research is the #1 use case of AI on Apollo - not messaging, not automated workflows. As Jeff Bezos says, "When the data and the anecdotes disagree, the anecdotes are usually right." In this case, they aligned perfectly. AI's real home in outbound isn't writing emails. It's list-building, account analysis, and qualification. Nobody wants "I see you went to Stanford" anymore. They want you to understand their actual problem. Check out the full case study: https://lnkd.in/gFdaaE9q

Rather than an AI assistant, productivity chatbots will increasingly look like a whole research team. An obvious path to continue improving the usefulness of LLMs is to increase test-time compute by running the same query in parallel against multiple models. Perplexity just released Model Council. It tasks three different models with researching user queries, followed by a synthesis phase where a fourth model aggregates findings to pinpoint insights and divergence. This allows users to directly compare advanced model performance within a consolidated interface.

For all the flak Perplexity gets because it's doing many things, I'm very impressed with their latest "Labs": Research++ Where previous research reports held themselves mostly to text and tables, Perplexity takes it a step forward by interjecting various computation steps, which allows it to take data from its sources and create visualisations on the fly. The result is a fully referenced, to-the-point report that takes about 10 minutes, which doesn't just give you the raw research, but also the analysis. It's for pro users only at the moment, but worth a shot!