Agents · Lesson 05 Pro+ ~11 min read Autonomous builds

Labs: agentic builds, not searches.

Labs is Perplexity's autonomous mode for long-running tasks. Unlike a search or a chat answer, Labs takes a prompt and produces a finished artifact — a working web app, a structured spreadsheet, a multi-page document, even a small site. It's the most ambitious of Perplexity's agentic modes and the one that hints at where AI is going.

Workflow 01 Build a small app from a prompt

1

Functional prototypes in 20 minutes

The headline Labs use case: describe an app, Labs builds it. Not 'show me how to build,' but 'here's the working app.'

The prompt that works

Tool buildBuild me an interactive web tool that: • Lets a user paste in their resume • Asks them to specify a job title and a job description (paste the JD) • Compares the resume to the JD and produces: - Top 5 skills present in both (good match) - Top 5 skills the JD requires that aren't on the resume (gaps) - 3 suggested resume bullet rewrites to better match the JD - A 1-10 match score with explanation UI: clean, minimal, dark mode. Mobile-friendly. No login required. Save as a shareable URL.

Best use cases

  • Internal tools for specific repetitive tasks
  • Customer-facing utilities (calculators, configurators)
  • Demo prototypes for stakeholder reviews
  • Personal productivity tools
Labs-generated apps are prototypes, not production-ready. For anything customer-facing or business-critical, treat the Labs output as a starting spec, not a finished product.
Time savings: From idea to interactive prototype: weeks of dev → an hour.

Workflow 02 Long-running research + deliverable

2

Hand Labs a multi-hour project

For research that produces a finished artifact and would take you days, Labs runs in the background and produces the deliverable.

The prompt that works

Multi-hour deliverableLong-running task: Research and produce: a comprehensive 25-page market analysis on the 'AI-native CRM' category, mid-2026. Sections: 1. Definition and scope 2. Market size and forecast (with charts) 3. Top 12 vendors (one page each: company, product, traction, funding, key customers, weaknesses) 4. Customer demand patterns (from interviews, surveys, analyst data) 5. Technology trends (what's changing, what's stable) 6. 18-month outlook 7. Strategic recommendations 8. Appendix: methodology + sources This will take 60-90 minutes to complete properly. Run it in the background. Save as both Word doc and PowerPoint outline. Cite every claim.

Best use cases

  • Industry analysis reports
  • Comprehensive vendor evaluations
  • Strategic deep-dives
  • Investor or board research packages
Long Labs runs need monitoring. Sometimes they go off-track and you only discover hours later. Check in at midpoint to verify direction.
Time savings: A comprehensive market report: days of consulting → 90 min of agent time.

Workflow 03 Combine Labs with Spaces

3

Ongoing projects that produce deliverables

The interesting Labs pattern: combine with a Space. The Space holds ongoing context and sources; Labs produces specific deliverables from the Space's accumulated knowledge.

The prompt that works

Space + LabsPattern: ongoing research + on-demand deliverables 1. Maintain a Space for [topic] — e.g., 'AI infrastructure landscape' 2. Add sources, custom instructions, and notes to the Space over time 3. When you need a deliverable (deck for a meeting, report for a stakeholder), invoke Labs from within the Space 4. Labs uses the Space's context as input → produces the specific deliverable you asked for Example invocations: • 'Build a 10-slide briefing for my board from this Space's research' • 'Build a one-pager I can send to a prospect summarizing this market' • 'Build an interactive comparison tool from the vendor data in this Space'

Best use cases

  • Ongoing competitive intelligence with on-demand briefings
  • Sales-enablement materials drawn from team knowledge
  • Stakeholder reports from accumulating research
  • Strategic memos pulled from continuous market watch
Labs deliverables are only as good as the Space's sources. Audit Space quality before generating high-stakes deliverables from it.
Time savings: Ongoing context + finished deliverables: the most-leveraged pattern.

Run one Labs task end-to-end

Pick a project you've been putting off because it'd take days. Frame it for Labs. Set it running. Come back to the deliverable. Notice the gap between 'I should build this someday' and 'it's built' shrinking to hours.

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

What you can do now

  • Use Labs for finished artifacts, not just answers
  • For long runs, check in at midpoint to confirm direction
  • Treat Labs apps and tools as prototypes, not production
  • Combine with Spaces for ongoing-research + on-demand-deliverable patterns
  • Verify data and claims before sharing Labs output externally
Pro+
Up next in Perplexity Mastery

The full Perplexity ecosystem

You've now covered Comet, Spaces, Deep Research, Model Council, and Labs — every major Perplexity agentic mode. The next horizon is integration: connecting Perplexity to your CRM, your data, your workflows. See the track →