Perplexity Chat

byPerplexity AI

Live-web research with numbered citations, so your moodboard, script or campaign starts from what is true this week

Start the research
Carved walnut gondola forcola marked R. Berto, photographed on grey, on a vermilion catalogue page with a cited note and its three source links

How Perplexity Chat works

Brief the analyst, scope one sharp question, choose how deep to dig. The answer arrives as a compact, cited block you can drop into the next node.

Perplexity Chat node set to sonar-pro with a documentary-desk system prompt, answering with a cited Guérande salt shot list

Set the analyst's role

Use the system prompt to define a design researcher, the output format, and rules like name sources inline and flag uncertainty.

Perplexity Chat node on sonar-pro with a scoped 30-day natural-dye question typed in, returning dated, cited themes

Scope the question

Write the prompt like a brief: a verb, a niche, a timeframe, and the shape you want back, such as five bullets or a hex palette.

Perplexity Chat node on sonar-reasoning-pro weighing Murray against Baedeker as inventor of the modern guidebook, with citations

Pick a tier and pipe it on

Choose sonar, sonar-pro or sonar-reasoning-pro, then send the cited answer to the next node on your canvas.

What Perplexity Chat is good at

Fresh facts, traceable sources and structured shapes. It is the research layer that tells you what is true before anything gets generated.

Indigo dossier page headed Weak signals: four numbered natural-dye trends with dates, citation numbers and the URLs they cite

Weak-signal trend scouting

Ask what is gaining traction in a niche over the past 6 months or 30 days and get compact, recency-aware answers with citation markers. A hunch becomes a moodboard-ready list of themes.

Three colour columns tearing down printed guidebooks, travel podcasts and short-form travel video, each with tone, a cited shift and source

Citation-ready competitor teardowns

Compare brand identity, tone and recent campaigns across rivals in a structured table. Every row points back to a source you can check before it reaches a client deck.

One guidebook question answered by sonar, sonar-pro and sonar-reasoning-pro in three blue columns, each with its own source list

Three tiers of depth

Use sonar for fast fact checks, sonar-pro for wider sourcing across a competitive landscape, and sonar-reasoning-pro when sources conflict and you need a weighed synthesis. Cap reply length to keep reasoning output tight.

Photo of a Soviet-era concrete bus shelter with a sun-and-wheat mosaic above a turquoise panel with its literal read and cited context

Reference images, grounded context

Attach a reference image and pair it with a specific question about what to look up. It identifies and contextualizes the subject, then adds live web research around it. Use a vision-focused model for deep compositional critique.

Made with Perplexity Chat

Every example here is a research job: trend lists, tone teardowns, reference shot lists and tags, each written to be pasted into a brief or a prompt.

Close crop of hands chiselling a zellige tile on a dusty bench, above the art direction note that produced the photo

Trend scouting condensed into an art direction note

Sleeper-train brief on midnight blue: a large 76%, two cited night-train stats and tone notes, above a lamplit compartment photo

Sourced stats and tone notes for brand copy

Long-exposure photo of a Breton fish market with ghosted figures beside the cited Autochrome finding and the prompt that made it

Research first, then image prompts for the canvas

Paludier skimming fleur de sel with a long-handled lousse, above a five-shot documentary list with one cited fact per shot

A shot list built from reported production facts

Black-and-white photo of a newspaper kiosk and a boy cycling past, below its alt text, five tags and its cited photographic tradition

Literal image description with web-backed tags

What people build with Perplexity Chat

Art directors, copywriters, filmmakers and archivists reach for it when the work depends on things that are current and checkable, and the final voice belongs to someone else.

Art directors building moodboards

01

Scout what is rising in a niche this month, gather named reference brands and campaigns, and walk into the moodboard with a cited brief instead of a hunch.

Copywriters needing fact-checked briefs

02

Pull key stats, claims and recent campaign context with sources attached, then hand the brief to a voicier writer or model. It supplies the facts and leaves the flair to the final author.

Canvas builders writing better prompts

03

Research the period, material or movement before you generate. Ask for descriptions grounded in real references, then pipe them to a stylistic model that turns them into polished image or video prompts.

Filmmakers prepping reference shot lists

04

Check period details, locations and reported production choices before breaking a scene down. Get a cited reference list that keeps the storyboard honest.

Librarians tagging visual archives

05

Attach an image and get a literal description, alt text and tags tied to the movement or era it belongs to, with web context cited for review.

Perplexity Chat specs

Plans and credits
Versions
sonar, sonar-pro, sonar-reasoning-pro
Inputs
Prompt (text), System Prompt (text)
Output
text

Perplexity Chat questions

It is best at web-grounded research: trend and competitor briefs, fact-checking claims before you script them, cited reference digging for moodboards, and current-events context for campaign copy. It runs on Perplexity AI's Sonar family, which retrieves live web pages and answers with numbered citations.

Yes. Sonar answers from a live web search and usually returns numbered citations with their source URLs. The numbers in the text don't always match the order of the source list, so open the sources that matter before a fact goes into a client brief.

Yes. Attach a still and it describes what is literally in the frame, then adds web-sourced context such as the place, period or tradition it belongs to, with citations you can check.

Match the tier to the question. sonar suits quick, cheap lookups. sonar-pro adds wider source coverage for multi-part research and competitive landscapes. sonar-reasoning-pro handles the hardest synthesis and conflicting sources, with slower, longer replies that may show its thinking.

Numbered markers like [1][2] point to the sources behind each claim, and sonar-reasoning-pro can prepend a <think> block before its answer. Both may need stripping before the text moves downstream. It can also paraphrase a source too confidently, so verify the claims that matter.

Answers come from a live web search, so they shift as sources change, and a source can occasionally be mis-summarised. Rerun the question when you need current figures, and open the citations that matter before a fact goes into a client brief.

Try Perplexity Chat on Fuser

Live-web research with numbered citations, so your moodboard, script or campaign starts from what is true this week

Start the research

Related models

Perplexity Chat docsAll models

OpenRouter Chat

OpenRouter (aggregator); underlying models vary by selection

ChatGPT

OpenAI

Google Gemini Chat

Google DeepMind

Groq Chat

OpenAI

Mistral Chat

Mistral AI

Claude Chat

Anthropic