Digital Digging × AI Training for TV 2 Journalists
20 Exercises for Perplexity
Every claim, every source, every link
April 2026 | 20 Exercises | Perplexity
Perplexity is unlike the other AI tools in this training. It is not a general-purpose language model pretending to know things — it is a search engine that reads the web in real time and cites its sources for every claim. That makes it uniquely valuable for journalists: you get answers with clickable references you can verify on the spot. Pro Search mode goes deeper, running multiple searches and synthesizing results into a structured brief. For fact-checking, source discovery, and quick background research, Perplexity is often the fastest path from question to verified answer.
But Perplexity has real weaknesses you need to understand. The sources it cites may be low-quality blogs, SEO spam, or outdated pages. It can misrepresent what a source actually says — quoting selectively or drawing conclusions the source does not support. The free tier has no document upload capability. And its analysis is shallow compared to Claude: it finds and summarizes, but it does not think deeply about contradictions, implications, or what is missing. Perplexity tells you what the web says. It does not tell you whether the web is right.
These 20 exercises are designed to build a systematic Perplexity workflow: how to ask, how to verify, how to catch the tool when it misrepresents its own sources, and how to combine it with other tools when Perplexity reaches its limits. Every exercise includes a source verification step — because with Perplexity, clicking the links is the whole point.
Exercises 1 – 7
Beginner
1
Source Check: One Question, Five Sources
Beginner
What you will learn
How to evaluate the quality and accuracy of Perplexity's source citations. You will discover that "cited" does not mean "correct" — and that the number of sources matters less than their reliability.
1.Ask Perplexity any factual question about Denmark. Something you already know the answer to, or can quickly verify. Example: "How many municipalities are there in Denmark and when was the last municipal reform?"
2.Count the sources cited in the response. Perplexity shows numbered references inline — note how many there are.
3.Click every single source link. For each, document: Does the page load? Is it from a credible source? Does it actually say what Perplexity claims it says?
5.Repeat with a more obscure question — something hyper-local or very recent. Compare source quality between the two queries.
What to watch for
Perplexity often cites Wikipedia, random blogs, and SEO-optimized pages alongside legitimate sources. A response with 8 citations may only have 2 that actually support the claim. The more obscure the question, the worse the source quality tends to be. This is your baseline: always click the links.
TV 2 connection: Peter — building systematic verification habits, starting with the most basic question: do the sources actually say what the AI claims?
2
Expert Finder
Beginner
What you will learn
How to use Perplexity to quickly identify domain experts in Denmark — and how to verify that the experts it suggests are actually leading researchers, not just people with a website.
1.Pick a topic relevant to a current story or beat. Ask Perplexity: "Who are the leading researchers on [topic] in Denmark? For each, give me their university affiliation, most cited publication, and public email address if available."
2.For each suggested expert, verify: Does this person exist at the named university? Check the university's staff page directly.
3.Check the "most cited publication" — search Google Scholar. Does the paper exist? Is it actually by this person? Is it their most cited work?
4.Verify email addresses. Are they real and current? (Do not send test emails — just check if the format matches the university's pattern.)
5.Ask yourself: are these genuinely the leading researchers, or just the most visible online? What names are missing that you know from your own reporting?
What to watch for
Perplexity tends to surface researchers who are active on social media or frequently quoted in press, not necessarily those with the strongest academic credentials. It may mix up affiliations (listing someone at KU who moved to AU years ago) or cite publications that do not exist. Always cross-check against Google Scholar and the university's own pages.
TV 2 connection: Marie — finding the right expert sources quickly for interview preparation and story context.
3
The Link Verification Drill
Beginner
What you will learn
The actual reliability rate of Perplexity's source links. By systematically checking 25 links across 5 questions, you will build an intuitive sense of how much you can trust the citations — and where the gaps appear.
1.Ask Perplexity 5 different factual questions — mix topics: one about Danish politics, one about science, one about a company, one about a recent event, one about history.
2.For each response, click every source link. Not just the first one — every single citation.
3.Score each link on two criteria: (a) Does the link work? (b) Does the source actually support the specific claim Perplexity attached it to?
4.Calculate your overall score: What percentage of links work? What percentage actually support the claims?
5.Look for patterns: Which types of questions produce the most reliable sources? Which produce the most broken links or misrepresented sources?
What to watch for
Expect a significant gap between "link works" and "link supports the claim." A link may load perfectly but say something subtly different from what Perplexity claims. Recent events tend to have worse source quality. Historical or well-documented topics tend to be more reliable. This exercise gives you your personal trust calibration.
TV 2 connection: Lars — building trust through evidence, not faith. This exercise turns the abstract question "Can I trust AI?" into a concrete, measurable answer.
4
Compare with Google
Beginner
What you will learn
Where Perplexity adds value over traditional search — and where Google still wins. Understanding the boundary between these tools prevents you from using Perplexity when Google would be faster, and vice versa.
1.Choose a research question you would normally Google. Something moderately complex — not a single fact, but a question requiring synthesis from multiple sources.
2.Search the same query in both Perplexity and Google. Time yourself: how long does it take to get a usable answer from each?
3.Compare results: What does Perplexity find that Google buries on page 3 or beyond? What does Google surface that Perplexity misses entirely?
4.Note the source overlap: How many of Perplexity's cited sources also appear in Google's first two pages of results?
5.Repeat with a very time-sensitive query (something from today). Which tool handles recency better?
What to watch for
Perplexity's advantage is synthesis — it reads multiple pages and combines the information. Google's advantage is breadth and recency. For very recent events, Google often wins because Perplexity's web index lags slightly. For complex background research, Perplexity can save significant time. Neither replaces the other.
TV 2 connection: Lars — settling the question of whether Perplexity actually beats Google, with evidence rather than assumptions.
5
Current Events Check
Beginner
What you will learn
How current Perplexity's information actually is — and where its real-time search hits a wall. You will discover the gap between "real-time" as marketed and "real-time" as experienced.
1.Ask Perplexity about something that happened today. A news event, a political statement, a sports result — something published within the last few hours.
2.Check: Does Perplexity have it? If yes, how current is the information? Check the timestamps on the sources it cites.
3.Now ask about something from yesterday. Then from last week. Map the reliability curve: at what age does information become consistently accurate?
4.Try a story that is still developing — where new facts are coming in throughout the day. Does Perplexity update its answer if you ask again an hour later?
5.Ask about a recent event you covered yourself. Does Perplexity's version match what you reported? Where does it diverge?
What to watch for
Perplexity's "real-time" search is faster than ChatGPT's web browsing but not instant. For breaking news within the first hour, it may have nothing or only early (often inaccurate) reports. For events 24+ hours old, it is usually reliable. The dangerous zone is 1-6 hours: Perplexity may have partial information presented with full confidence.
TV 2 connection: Sebastian — understanding the practical limits of "real-time" AI search for daily news work.
6
The Simple Fact Finder
Beginner
What you will learn
How accurate Perplexity is on simple, verifiable facts — and whether it can handle temporal comparisons (now vs. then) without confusing dates, numbers, or sources.
1.Ask Perplexity: "What is the current population of [Danish city]? Source?" Choose a city you can verify — Odense, Aarhus, or a smaller municipality.
2.Verify the answer against Danmarks Statistik (dst.dk). Is the number correct? Is the source Perplexity cites the actual origin of the data?
3.Now ask: "What was the population of [same city] 10 years ago? Source?"
4.Verify again. Does the historical figure match DST records? Does Perplexity cite a legitimate source for the historical data, or is it pulling from a random website?
5.Ask Perplexity to calculate the difference and percentage change. Check its math. Simple arithmetic errors in AI output are more common than you think.
What to watch for
Perplexity may give a population figure from 2023 while claiming it is "current." Historical figures are more error-prone — it may cite a source that uses a different municipal boundary (pre-2007 reform vs. post-reform). Always check whether the numbers are for the municipality or the city proper. And yes, AI tools regularly get basic arithmetic wrong.
TV 2 connection: Lasse — developing the habit of fact-checking even the simplest claims, because errors in basic facts undermine everything built on top of them.
7
Organization Background
Beginner
What you will learn
How to use Perplexity for rapid background research on organizations — and how to catch the tool when it presents incomplete or misleading background information.
1.Choose an organization relevant to your beat — a Danish NGO, government agency, or company. Ask Perplexity: "What is [organization]? When was it founded? Who are the key people? Have there been any controversies?"
2.Click every source link in the response. For each claim (founding date, key people, controversies), verify against the cited source.
3.Check the organization's own website. Does Perplexity's summary match the official information? Where does it diverge?
4.Look specifically at the "controversies" section. Are these real? Are they current? Or is Perplexity surfacing old, resolved issues as if they are still active?
5.Search the organization in a Danish news archive (Infomedia or similar). What does Perplexity miss that actual journalism has covered?
What to watch for
Perplexity's "controversies" section is especially unreliable. It may surface a single blog post as a "controversy" or miss a major scandal entirely because Danish-language coverage was not indexed. Key people may be outdated (someone who left two years ago). Founding dates are often wrong for organizations that have been restructured or renamed.
TV 2 connection: Joachim — using Perplexity for OSINT-style background research while maintaining the discipline of verification.
Exercises 8 – 14
Intermediate
8
The OSINT People Search
Intermediate
What you will learn
How Perplexity performs as an OSINT tool for gathering publicly available information about individuals — and where it falls short compared to manual search techniques.
1.Choose a public figure — a politician, company CEO, or public official. Ask Perplexity: "What public information is available about [name]? Include their professional history, published statements, organizational affiliations, and any public records."
2.Verify each claim: Check LinkedIn (if public), company registries (CVR), and news archives. What does Perplexity get right? What does it get wrong?
3.Now do the same search manually: Google the name, check social media, look at CVR, search Infomedia. What information do you find that Perplexity missed?
4.Compare the two approaches: Which was faster? Which was more comprehensive? Which was more accurate?
5.Note specifically: Does Perplexity access Danish-language sources as effectively as English-language ones? Is the Danish coverage thinner?
What to watch for
Perplexity is faster than manual OSINT for getting a first overview, but it consistently misses Danish-language sources, paywalled content, and information behind social media logins. It may also confuse people with similar names — especially common Danish names. Never trust a Perplexity people search without manual verification of every fact.
TV 2 connection: Jakob and Nanna — comparing AI-assisted people search against manual OSINT techniques like Facebook research and registry lookups.
9
Criminal Network Research
Intermediate
What you will learn
How Perplexity handles sensitive investigative queries — specifically, whether it can accurately compile publicly reported information about criminal organizations while properly attributing each claim to a specific source.
1.Ask Perplexity: "What has been publicly reported about [known criminal organization] in Denmark in the past 2 years? Provide sources for each claim." Use an organization that has been widely covered in Danish media.
2.For each claim in the response, check: Is this sourced to actual journalism? Or is it pulled from a blog, a forum, or a low-quality aggregator?
3.Verify the chronology: Does Perplexity accurately report when events happened? Are any of the "past 2 years" claims actually older?
4.Check for mixing: Does Perplexity conflate different organizations, incidents, or individuals? This is common in crime reporting where names and events are similar.
5.Compare against your own knowledge or a colleague's: What major developments does Perplexity miss? What does it include that was not widely reported?
What to watch for
Criminal journalism is high-stakes — errors can have legal consequences. Perplexity may cite sources that have since been corrected or retracted. It may present allegations as facts or mix up individuals involved in different cases. It also struggles with Danish-language crime reporting because much of it is behind paywalls. Use this only as a starting point for your own research, never as source material.
TV 2 connection: Franziska — testing Perplexity's usefulness for criminal journalism research while understanding its severe limitations in this sensitive domain.
10
Legislation Tracker
Intermediate
What you will learn
How well Perplexity tracks legislative changes — whether it can accurately identify the status, sponsors, and public reactions to proposed or enacted Danish laws.
1.Ask Perplexity: "What changes to [Danish law] have been proposed or enacted in 2025-2026? For each change, give me the status, who proposed it, and public reactions. Source each claim."
2.Click every source. For legislative status, verify against ft.dk (Folketinget's own website). Is the status Perplexity reports actually current?
3.Check the "sponsors" — does Perplexity correctly identify who proposed the legislation? Verify against the official parliamentary records.
4.Evaluate the "reactions" section: Are these from actual stakeholders, or is Perplexity pulling generic commentary? Are the quoted reactions accurately attributed?
5.Ask a follow-up: "What is the expected timeline for [specific proposed change]?" Verify this against parliamentary scheduling.
What to watch for
Legislative tracking is where Perplexity's real-time search should shine — but Danish parliamentary processes are not always well-indexed in English-language sources. Perplexity may conflate proposals with enacted laws, or report on bills that were withdrawn. Always cross-check against ft.dk, which is the definitive source for Danish legislation status.
TV 2 connection: Pelle — tracking regulatory changes that affect data security, AI governance, and journalism practice, with verified sources at every step.
11
The Contradiction Checker
Intermediate
What you will learn
Whether Perplexity gives you the same answer twice — and what it means when it does not. Consistency is a proxy for reliability, and this exercise measures it directly.
1.Ask Perplexity a moderately complex factual question. Screenshot the response and save all cited sources.
2.Wait at least 24 hours. Ask the exact same question again. Screenshot this response too.
3.Compare: Do the sources change? Are different websites cited? Does the answer itself change — different numbers, different emphasis, different conclusion?
4.Document every difference. For each change, determine: Is the new answer better (updated information) or just different (inconsistency)?
5.Try a third time a week later. Map the drift: Is Perplexity converging on a stable answer, or does it fluctuate randomly?
What to watch for
Perplexity's answers change because the web changes — new pages get indexed, old ones disappear, and the search algorithm may weight sources differently on different days. This is not necessarily a flaw, but it means you cannot treat a Perplexity response as a stable reference. If you cite a Perplexity finding, go to the underlying source directly and cite that instead.
TV 2 connection: Lars — quantifying the trust question. If the same tool gives different answers to the same question, what does that mean for its reliability?
12
International Coverage Comparison
Intermediate
What you will learn
How to use Perplexity to discover what foreign media reports about Danish stories — specifically, facts or angles that exist in German, Swedish, or British coverage but are absent from Danish reporting.
1.Choose a Danish story with international relevance. Ask Perplexity: "How has [Danish story] been covered in German media? What facts are reported in German coverage that are missing from Danish reporting?"
2.Repeat for Swedish media and UK media. Note the differences in framing and factual content across each country's coverage.
3.Verify: Click the international sources Perplexity cites. Do they actually exist? Do they actually report the facts Perplexity claims?
4.Identify discrepancies: Where do international reports disagree with Danish coverage? Are these genuine factual differences, or differences in emphasis and framing?
5.Ask: "Are there any facts reported by foreign media that no Danish outlet has covered?" Verify any such claims independently.
What to watch for
Perplexity's multilingual search is better than most AI tools, but it still favors English-language sources. German and Swedish sources may be underrepresented. It may also attribute information to a foreign source that actually originated in Danish media and was merely translated. Trace the chain: where did the information first appear?
TV 2 connection: Pelle and Mads B. — using Perplexity's web search to compare international coverage and find angles that Danish media may have missed.
13
The Academic Source Finder
Intermediate
What you will learn
How well Perplexity finds and accurately represents peer-reviewed research — and how to catch it when it fabricates or misattributes academic publications.
1.Ask Perplexity: "What are the key peer-reviewed studies about [topic] published in the past 3 years? For each, give me the authors, journal, main finding, and methodology."
2.For each study cited, search Google Scholar or PubMed. Does the paper exist? Are the authors correct? Is it published in the journal Perplexity claims?
3.Read the abstracts of the papers that actually exist. Does Perplexity accurately represent the findings? Or does it oversimplify, exaggerate, or misstate the conclusions?
4.Check the "past 3 years" constraint: Are all cited papers actually from 2023-2026? Perplexity often includes older papers while claiming they are recent.
5.Ask a follow-up: "Are there any meta-analyses or systematic reviews on this topic?" Verify these exist — meta-analyses are particularly prone to fabrication.
What to watch for
Perplexity is better than ChatGPT at academic searches because it actually searches the web, but it still fabricates papers — particularly when asked about niche topics. It may cite real journals with fake article titles, or real authors with papers they never wrote. The methodology descriptions are often generic rather than specific to the actual study. Always verify in Google Scholar before citing.
TV 2 connection: Peter — finding scientific sources for verification work and understanding how to catch AI when it fabricates academic credentials.
14
Company Due Diligence
Intermediate
What you will learn
How to use Perplexity for rapid company background research — ownership structures, financial data, legal issues — and where Perplexity's publicly available web data runs out.
1.Choose a Danish company. Ask Perplexity: "What is publicly known about [company]? Ownership structure, revenue, number of employees, any legal issues, regulatory actions, or controversies."
2.Verify ownership against CVR (virk.dk). Does Perplexity correctly identify the owners, parent companies, and subsidiaries?
3.Check financial data against the company's most recent annual report (available on virk.dk for most Danish companies). Are the numbers correct and current?
4.Search for legal issues independently: Check domstol.dk, news archives, and the Danish Business Authority. What does Perplexity miss?
5.Ask a follow-up: "What companies are connected to [company] through shared board members or ownership?" Verify the network Perplexity describes.
What to watch for
Perplexity gives you a useful first sketch but misses critical details. Danish company registrations, ownership chains, and legal filings require direct access to CVR and court records — data that Perplexity cannot always access or index. Financial figures may be from the wrong year. Board member lists may be outdated. Use Perplexity to generate leads, then verify everything through primary sources.
TV 2 connection: Mathias — rapid company research for investigations, understanding both the power and the limits of AI-assisted due diligence.
Exercises 15 – 20
Advanced
15
Deep Dive: Danish Media Landscape
Advanced
What you will learn
How to use Perplexity for structural research on media ownership — testing its ability to map complex corporate structures, board memberships, and cross-ownership patterns in an area you know well enough to verify.
1.Ask Perplexity: "What are the ownership structures of the 5 largest Danish media companies? For each, identify the parent company, major shareholders, board members, and any cross-ownership with other media companies."
2.Verify every claim. Check CVR for ownership structures. Check the companies' own annual reports for board membership. Cross-reference with Slots- og Kulturstyrelsen's media ownership register.
3.Map the connections Perplexity identifies. Are the cross-ownership claims accurate? Or does it confuse historical ownership with current structures?
4.Ask: "Have any of these ownership structures changed in the past year?" Verify any claimed changes against news reporting and registry filings.
5.Identify the gaps: What connections does Perplexity miss that you know exist from your industry knowledge? What would it take to build a truly complete media ownership map?
What to watch for
Media ownership in Denmark is complex — foundations (fonde), cross-holdings, and international parent companies create layers that Perplexity struggles to untangle. It may present outdated ownership structures (pre-merger or pre-sale) as current. Board member lists are especially prone to errors because they change frequently and are not always well-indexed online.
TV 2 connection: Christian — applying critical media analysis skills to AI output about the very industry you work in. If Perplexity gets your own industry wrong, what does that mean for its accuracy on industries you know less about?
16
Real-Time Verification Drill
Advanced
What you will learn
How to use Perplexity as a verification tool during breaking news — specifically, how to trace claims back to original sources, find contradicting reports, and distinguish independent reporting from circular citation.
1.During a breaking news event (or reconstruct a recent one), take a specific claim circulating in the media. Ask Perplexity: "What is the original source for the claim that [specific claim]? Trace it back to where it first appeared."
2.Verify the source chain Perplexity provides. Does it actually trace back to the origin, or does it just find the earliest indexed version?
3.Ask: "What reports contradict or complicate this claim? Provide sources." Click every link. Are these genuine counterpoints or tangential articles?
4.Now the critical question: "Of the media outlets reporting this claim, which ones cite the same original source and which conducted independent reporting?" Verify Perplexity's assessment.
5.Document the full verification chain. How long did this take with Perplexity compared to manual verification? Where did Perplexity add value, and where did it waste your time?
What to watch for
Perplexity struggles to distinguish between original reporting and citation chains. It may identify 5 "independent sources" that all trace back to a single wire service dispatch. During breaking news, it may also present early (incorrect) reports alongside later corrections without flagging the discrepancy. Speed is Perplexity's advantage here — but speed without accuracy is worse than slow verification.
TV 2 connection: Pelle and Mads B. — building a breaking news verification workflow that uses Perplexity's speed while maintaining journalistic rigor.
17
The Source Tree
Advanced
What you will learn
How to trace a claim through multiple layers of sourcing — using Perplexity to dig beneath the surface source to find where information actually originates, and where the chain breaks.
1.Take a specific news claim — something stated as fact in a recent article. Ask Perplexity: "What is the original source for the claim that [claim]?"
2.Whatever source Perplexity identifies, ask: "And what is THAT source's source? Where did they get this information?"
3.Go one more level deep: "And where did THAT information originate? What is the primary source — the first document, statement, or data set?"
4.Verify the chain independently. At each level, click the source and check: Does Perplexity correctly identify the sourcing? Or does it lose the thread?
5.Draw the source tree on paper. Where does the chain end? Is the ultimate source a press release? A study? A social media post? An unnamed official? What does this tell you about the reliability of the original claim?
What to watch for
Perplexity usually traces one or two levels deep before it starts guessing or conflating sources. It may identify a secondary source as "the original" because that is the oldest result in its index. The deeper you go, the less reliable Perplexity becomes — which is exactly where the most important information lives. This exercise teaches you to recognize when you have hit the limit of what AI search can do and need to switch to manual investigation.
TV 2 connection: Peter — understanding the source chain is the foundation of verification. This exercise builds the habit of never stopping at the first source.
18
Cross-Border Investigation
Advanced
What you will learn
How to use Perplexity for cross-border research — finding what is reported about a Danish entity in foreign jurisdictions, where coverage may be more revealing or less constrained.
1.Take a Danish company with international operations. Ask Perplexity: "What has been reported about [company] in Germany, the UK, and the US? Compare the coverage across these three countries."
2.Click every source. Are the foreign sources legitimate media outlets or industry publications? Are the reports actually about the Danish company, or about a similarly named entity?
3.Identify discrepancies: Where do the reports from different countries disagree? Are there facts reported abroad that have not been covered in Danish media?
4.Ask Perplexity: "Has [company] faced any regulatory actions, lawsuits, or investigations in any country outside Denmark?" Verify each claim.
5.Assess: How much of the cross-border picture could you have found without Perplexity? What would you have missed? What did Perplexity miss that a targeted search in each country's language would find?
What to watch for
Perplexity searches primarily English-language sources, even when asked about German or other foreign coverage. It may miss important reports in non-English media. It can also confuse subsidiaries with parent companies or mix up entities with similar names across borders. For genuine cross-border investigation, Perplexity is a starting point — you still need to search in each country's language and check local registries.
TV 2 connection: Mads B. and Pelle — using Perplexity's multilingual capabilities for cross-border investigative research while understanding the language bias in its results.
19
The Monitoring Brief
Advanced
What you will learn
How to use Perplexity to design and test a systematic monitoring workflow for your beat — and whether AI-assisted monitoring can reliably replace or supplement manual scanning.
1.Ask Perplexity: "Create a weekly monitoring brief template for a journalist covering [your beat]. What sources should be checked weekly? What keywords should be tracked? What patterns should be watched for?"
2.Evaluate the template: Are the suggested sources relevant and comprehensive? Are there sources you would add that Perplexity missed? Are there suggested sources that are not useful for your specific beat?
3.Test the template for one week. Every day, run the key queries from the template through Perplexity. Document what it surfaces.
4.Compare: What stories or developments did your Perplexity monitoring catch that you would otherwise have missed? What did it miss that your normal monitoring caught?
5.Refine: Based on the week-long test, revise the template. Which queries were useful? Which were noise? What would you add for the second week?
What to watch for
Perplexity's monitoring suggestions tend to be generic. A "crime beat" template will suggest obvious sources (police reports, court filings) but miss beat-specific angles (community forums, local council agendas that hint at upcoming enforcement actions). The value is in customization — start with Perplexity's template, then ruthlessly edit based on your own expertise. Also note: Perplexity has no alerting function, so monitoring requires manual daily queries.
TV 2 connection: Sanne and Franziska — building a practical monitoring system that uses AI search to augment (not replace) manual beat scanning.
20
Build Your Perplexity Research Workflow
Advanced
What you will learn
How to design a complete, reusable research workflow — from initial question to verified answer — that integrates Perplexity with other tools and manual verification at each step.
1.Based on everything you have learned in exercises 1-19, design a step-by-step research workflow. Write it down: (a) What questions do you ask Perplexity first? (b) In what order? (c) At which point do you verify sources? (d) When do you switch to another tool (Google, Claude, manual search)?
2.Test your workflow on a real story you are currently working on. Follow your own steps strictly — no shortcuts.
3.Document: Where did the workflow save time? Where did it add unnecessary steps? Where did Perplexity fail and you needed a different tool?
4.Revise your workflow based on the test. Remove steps that added no value. Add steps you realized were missing. Make the decision points explicit: "If Perplexity's sources are low-quality, switch to [X]. If the topic requires deep analysis, move to Claude."
5.Share your workflow with a colleague. Have them test it on a different story. Incorporate their feedback into a final version you can use daily.
What to watch for
The most common mistake is making the workflow too rigid. Every story is different — your workflow should have decision points ("if X, then do Y; if not, do Z") rather than a fixed sequence. The second most common mistake is skipping verification steps under time pressure. Build verification into the workflow as a non-negotiable step, not an optional add-on. The best workflow is one you actually use — if it is too complex, you will abandon it.
TV 2 connection: Emil and Lars — synthesizing everything from this training into a practical, daily-use system that makes Perplexity genuinely useful rather than occasionally interesting.
When to Use Perplexity — and When Not To
Use Perplexity when you need:
+Quick background research — organization histories, public figures, company basics. Perplexity synthesizes multiple sources faster than manual Google searches.
+Source discovery — finding who has reported on a topic, what experts exist, which publications cover a subject. The inline citations are genuinely useful for building source lists.
+Current events overview — what has been reported in the last 24-48 hours, from multiple outlets. Faster than scanning individual news sites.
+International coverage — what foreign media says about a Danish story. Perplexity's web search covers multiple languages better than ChatGPT.
+Fact-checking starting points — "is this claim true?" with sources you can immediately verify. The citations make verification efficient.
Do NOT use Perplexity when you need:
–Deep document analysis — Perplexity cannot upload and analyze PDFs, reports, or legal documents (free tier). Use Claude instead.
–Complex reasoning or critique — Perplexity finds and summarizes. It does not think critically about contradictions, implications, or unstated assumptions. Use Claude for analysis.
–Writing assistance — Perplexity is a search tool, not a writing partner. Use ChatGPT or Claude for drafting, editing, and style work.
–Paywalled or proprietary information — Perplexity only searches the open web. It cannot access Infomedia, court databases, or subscription-only sources.
–Definitive verification — Perplexity is a starting point, never the endpoint. Its sources may be low-quality, outdated, or misrepresented. Always go to the primary source.