Research with Luma
How Research Works in Luma AI
The Luma Agent has web research capabilities accessible across both Brainstorm Mode and Create Mode. Research isn't a separate tool — it's woven into the conversational workflow. Ask a question, and Luma searches the web in real time, retrieves relevant sources, synthesizes the findings, and presents them directly in the conversation or as a structured document on the board.
The output is curated information — not a list of links. Sources are read, key data is extracted, and findings are organized for immediate creative use. Research can inform ideation, validate creative direction, or provide technical specifications mid-production.
Where Research Lives
Brainstorm Mode — Research is most commonly used here during the ideation phase. Ask for competitor analysis, trend reports, visual references, or cultural context to shape creative direction before any assets are generated.
Create Mode — Research is also available during production. Need to check a platform's aspect ratio specs mid-build? Want to reference a specific film's color grading approach while generating images? Ask during creation and the information feeds directly into the workflow.
Contextual Memory — Any asset already on the board — images, videos, documents, brand guidelines — can be analyzed and referenced during research. Upload a competitor's ad and ask how to differentiate. Drop in brand guidelines and ask for research that aligns with them.
Research Best Practices
Be Specific About What's Needed
Frame requests around the actual information gap, not the search mechanics.
The more precisely the question is framed, the more actionable the output.
Layer Questions — Don't Front-Load
Research works best as a conversation, not a single megaprompt. Start broad, then narrow based on what comes back.
Round 1: "What visual styles are trending in luxury real estate marketing?" Round 2: "Focus on the minimal editorial direction — who's doing it best?" Round 3: "Pull specific examples from those brands I can reference for my campaign."
Each round builds context. This layered approach produces sharper, more targeted results than trying to ask everything at once.
Use Board Assets as Research Context
Drop reference images, competitor screenshots, brand guidelines, or existing work onto the board before starting research. Luma can analyze what's already there and tailor findings accordingly.
Effective combinations:
Brand guidelines PDF + research request → Results filtered through brand parameters
Competitor screenshot + differentiation question → Analysis grounded in specifics
Existing campaign asset + trend research → Recommendations that build on current work
Product photo + market research → Positioning insights tied to the actual product
Ask for Structured Deliverables
When research feeds a project, specify the output format:
"Summarize this into a competitive analysis document"
"Create a reference sheet with key takeaways and source links"
"Organize findings into a mood direction brief"
"Build a platform specs cheat sheet I can reference during production"
"Write a creative brief based on what we've found"
This produces a persistent asset — editable, referenceable, and available throughout the project lifecycle.
State the Purpose
Context about the end goal sharpens everything. There's a meaningful difference between research for inspiration and research for a deliverable.
For ideation: "I'm exploring visual directions for a rebrand — show me what's out there."
For production: "I'm building hero images for a sneaker campaign targeting 18–24 year olds. I need specific reference examples to guide the generation."
For validation: "I've chosen a brutalist design direction for this brand. Find examples of other premium brands that have used brutalism successfully."
For technical specs: "I'm creating video ads for TikTok, Instagram Reels, and YouTube Shorts. What are the current recommended specs for each?"
Purpose-driven requests produce research that's already shaped for its destination.
Research During Production — Not Just Planning
Research isn't limited to the brainstorming phase. During asset creation in Create Mode, research can answer questions in real time:
"What's the exact hex code for Coca-Cola red?"
"Show me examples of split-screen compositions in fashion photography"
"What film used the teal-and-orange grading I'm trying to replicate?"
"What's the standard safe zone for text in Instagram Stories?"
This eliminates tab-switching and keeps the creative flow unbroken.
What Research Can Cover
Visual trends — current styles, color palettes, typography, layout patterns across industries
Competitor analysis — what other brands are doing, how they position visually and strategically
Cultural context — memes, movements, subcultures, and aesthetics relevant to a target audience
Platform specifications — format requirements, best practices, and performance benchmarks for TikTok, Instagram, YouTube, LinkedIn, and more
Technical references — aspect ratios, resolution standards, duration recommendations, codec specs
Art and film references — specific directors, cinematographers, photographers, designers, movements — as creative vocabulary
Market research — audience demographics, positioning strategies, pricing context
Inspiration gathering — reference images, mood boards, stylistic touchpoints from across the web
Research Limitations
Paywalled content — research works with publicly available sources only
Real-time data precision — trending topics and statistics may have slight lag
Domain expertise depth — synthesizes available online information; highly specialized or proprietary industry knowledge may require human input
Asset generation — research informs decisions; image, video, and audio creation is handled by the generation tools
Historical accuracy — always verify critical dates, statistics, or claims against authoritative sources for high-stakes projects
Research Workflows by Use Case
For Brand Campaigns
Upload brand assets and guidelines to the board
Ask for competitor landscape analysis in Brainstorm Mode
Request trend research filtered through brand parameters
Generate a creative brief document from findings
Switch to Create Mode with the brief as a reference
For Content Creation
Describe the content format and platform
Ask for current best practices and specs for that platform
Request visual references that match the intended style
Generate assets in Create Mode, referencing research mid-production as needed
For Product Launches
Upload product photos to the board
Research competitor visual strategies in the same category
Ask for trend analysis relevant to the target demographic
Generate multiple creative direction options
Research which directions align with current market momentum
For Pitch Decks and Presentations
Describe the audience and objective
Research market data, competitor positioning, and visual benchmarks
Request a structured research summary document
Use findings as the data backbone for slide content and visuals
Quick Reference
One topic per thread — mixing unrelated research topics dilutes quality
React to what comes back — "more like this" or "not that direction" refines results faster than re-prompting from scratch
Ask for comparisons — "How does Brand A differ from Brand B?" beats researching each separately
Name the audience — stating target demographics upfront improves relevance across every finding
Save everything — request a summary document for any research that matters beyond the current conversation
Cross-reference modes — start research in Brainstorm, continue it in Create when production questions arise
Use research as generation context — reference research documents and findings when prompting asset generation for more targeted outputs