Agents · Lesson 03 Pro ~11 min read Reports + presentations

Deep Research: analyst output, on demand.

Perplexity Deep Research is the long-running research mode — now upgraded to state-of-the-art performance running on Claude Opus 4.5 for Max and Pro users. Unlike a chat answer, Deep Research can produce presentations, spreadsheets, dashboards, and websites directly as deliverables.

Workflow 01 Research → presentation deliverable

1

A finished deck, not a research report

The breakthrough Perplexity Deep Research feature: it produces presentation-ready output, not just text. Ask for slides; get slides.

The prompt that works

Slide-deck outputResearch: 'The state of vertical AI in legal tech, mid-2026.' Produce as: a 12-slide presentation deck (PowerPoint or web) covering: • Slide 1: Title + your name + date • Slide 2: Executive summary (3 key takeaways) • Slide 3-4: Market sizing and growth trajectory • Slide 5-7: Top 5 players — what they do, traction, funding • Slide 8-9: Use cases that are working in production • Slide 10: Unsolved problems / gaps in the market • Slide 11: 18-month outlook • Slide 12: Sources Use a clean dark-mode design. No stock photos. Cite sources on each data slide.

Best use cases

  • Investor pitch supporting research
  • Internal market briefings
  • Conference talk preparation
  • Board-level industry updates
Generated decks need a polish pass — your specific company language, brand colors, and voice. Treat as 80% there, not 100%.
Time savings: 12-slide research deck: 1-2 days work → 30 min + polish.

Workflow 02 Research → spreadsheet deliverable

2

Structured data, not just narrative

For research that produces data (comparison tables, market data, taxonomies), ask Deep Research for a spreadsheet output.

The prompt that works

Spreadsheet outputResearch task: comparison of 20 AI agent platforms for enterprises. Produce as: an Excel spreadsheet with these columns: • Platform name • Vendor • Year founded • Total funding raised • Primary use case category • Supported LLM providers (OpenAI, Anthropic, Google, etc.) • Self-hosted option? (Y/N) • Pricing model (per-seat, usage-based, enterprise-only) • Notable customers (3-5 examples each) • Key differentiator (one sentence) • Source URL for the data One row per platform. Multiple sheets if needed for related data (e.g., a separate sheet for pricing details).

Best use cases

  • Vendor evaluation matrices
  • Market mapping (competitive landscapes)
  • Multi-attribute comparison tables
  • Time-series data extracted from sources
Spot-check cells against original sources, especially funding numbers, customer claims, and pricing. AI tools sometimes confidently report outdated or wrong numbers.
Time savings: Vendor comparison spreadsheet: days → 30 min.

Workflow 03 Research → interactive dashboard

3

Live data exploration, not static report

Perplexity's newer capability: produce interactive dashboards from research. Useful when stakeholders want to explore data themselves.

The prompt that works

Dashboard outputResearch task: 5-year trend analysis of AI adoption by industry, US market. Produce as: an interactive dashboard with: • Industry filter (dropdown) • Year-over-year line charts (adoption %, spending) • Industry breakdown bar chart • 'Drill down' on any industry to see top use cases • Source citation panel that updates with each view Design: • Clean, business-appropriate (not flashy) • Mobile-responsive • Save as a shareable URL I can send to stakeholders

Best use cases

  • Executive briefings stakeholders can self-serve
  • Customer-facing data presentations
  • Investor data rooms
  • Ongoing tracking dashboards (refreshable)
Interactive dashboards are great for exploration but harder to fact-check. Make sure the underlying data is solid before publishing.
Time savings: Interactive dashboard from data: weeks of dev → an afternoon.
Deep Research vs. Spaces

Deep Research is the long-running mode for one-shot deep dives. Spaces are persistent workspaces for ongoing research. Use Deep Research for 'I need a finished report now.' Use a Space for 'I'll be researching this topic over weeks.'

Order a deliverable instead of a search

Pick a topic where you'd normally produce a slide deck or spreadsheet. Have Deep Research produce the deliverable. Polish the output. Compare to your manual workflow — including time, source quality, and design effort.

Back to Perplexity Mastery → Try a prompt in the playground →

What you can do now

  • Ask for specific output formats (deck, sheet, dashboard) — not just text
  • Verify data and numbers against original sources
  • Always polish output for voice, brand, and accuracy
  • Use Deep Research for one-shot deliverables, Spaces for ongoing work
  • Cite sources prominently in finished deliverables
Pro+
Up next in Perplexity Mastery

Lesson 04 · Model Council — compare answers from frontier models

Model Council is Perplexity's newest agent: ask the same question to multiple frontier models simultaneously and see how they differ. The pattern that catches blind spots. See the track →