AIAP Tool Guides·Perplexity·17:07

Perplexity Discover, Personalisation, Multimodal Uploads and Voice: Context About You, Image Analysis, Dictation and Two-Way Voice

Karan Rana Brand/marketing manager, YouTuber, podcaster and course creator ("a non-tech guy"); presents the AI for Marketing, NotebookLM and Manus tool guides

▶LMS lesson

The short version

Discover is an inspiration feed by topic; Personalization in settings stores a description of you, location and search-library use so answers fit you (0:00-0:06).
Attach an image: a rival cat-food poster dissected for design and copy with actionable counter-ideas; a stage photo read for body language (0:06-0:13).
Dictation types your speech (it misheard "reel hook" as "book"); Voice mode talks back - useful on commutes and runs (0:13-0:16).

At a glance, three clicks deep

Skim here first: the closed row is the glance, open is the study card with the key points and timestamps, and the ↓ link drops to that concept's full write-up below.

01Personalisation: tell it who you are - and weigh the privacyWrite about ten lines about yourself, your audience and interests in Personalization;›

Write about ten lines about yourself, your audience and interests in Personalization; optionally share location and let it reference your search library.

Discover (0:01-0:03)

Personalization (0:03-0:06)

↓ Full write-up of this concept

02Image analysis, dictation and two-way voiceAttach files (images, PDFs, audio, code) to a question;›

Attach files (images, PDFs, audio, code) to a question; use the mic to dictate, or Voice mode to converse.

Multimodality (0:06-0:08)

Poster and photo (0:08-0:13)

Dictation and voice (0:13-0:16)

↓ Full write-up of this concept

The concepts in full

Personalisation: tell it who you are - and weigh the privacy

how-to

Those lines "are what will make it highly contextual for you."

Settings > Personalization. He acknowledges real concerns - training on user data, sharing with third parties - and proceeds on the assumption that "trust boundaries are well respected", without stating Perplexity's policy. Check the current privacy settings before writing personal details.

Do it in this order

01Settings > Personalization: write a short profile.
02Decide on location and search-library access.
03Discover > personalize: choose topics.

Why it matters

Context once, better answers always - at a privacy cost to judge.

Image analysis, dictation and two-way voice

An analysis that would take "two, three hours" by hand.

Prompt: "this is a poster from my rival cat food brand - dissect and analyze it from a designer and copywriting POV, then advise how my brand's poster can stand out; concise bullet points." It covered layout, palette, the kitten's emotional pull, alliteration, vet trust and CTA, then suggested an unexpected colour pop, playful shapes, humour and verbs like "give your cat the wild nutrition they crave". Dictation garbled his request (partly into Devanagari) with an external mic; Voice mode's reply was not captured. Only images were demonstrated.

Why it matters

Multimodal input is how real work arrives.

This idea elsewhereparallelGemini: the same competitor-poster analysis

Tools referenced

Tool	Coverage	Moment	Context
Perplexity	demonstrated
Perplexity Voice Mode	demonstrated
Pinterest	mentioned

Action items

Resources mentioned

Resources

docLesson page on the platform

Extraction notes

This page was built from an auto-generated transcript, which garbles product and people's names. Those were corrected silently in everything above and logged here for transparency. The warnings flag claims that were true on the recording day but change fast.

Transcript corrections applied

The transcript says	The trainer actually means
real hook ideas	reel hook ideas
viscous	whiskers (likely)

True on recording day — verify before relying

Vendor UI walkthrough: menu names, layout, plan limits and pricing change monthly - check the live product before following click paths.
Discover categories, Personalization fields and the arrival of dictation and voice on the web app ("not there till last year") are mid-2025 (0:02-0:13).