Lookt Documentation
Learn how Lookt tracks your brand across AI engines, what the metrics mean, and how to act on the data.
Set up your workspace, add prompts, and run your first analysis in under 5 minutes.
Workspaces, prompts, brand mentions, source citations — the building blocks.
Dashboard, prompts, brands, sources, actions, and Lookt AI — what each does.
Visibility, sentiment, position, recommendation rate — what each number means.
What is Lookt
Lookt monitors how AI engines like ChatGPT, Google Gemini, and Perplexity mention your brand. Every day, we run your tracking prompts across these models and analyze the responses to measure:
Think of it as SEO for AI. Instead of tracking Google search rankings, you're tracking how AI assistants recommend and describe your brand to millions of users.
Quickstart
Get set up in under 5 minutes.
Sign up and create a workspace
Go to app.trylookt.com and sign up. You'll be guided through creating your first workspace.
Add your brand
Enter your brand name and domain. Lookt uses this to identify mentions in AI responses. You can add aliases too (e.g., "HubSpot" and "hubspot.com").
Set up tracking prompts
Prompts are the questions Lookt asks AI engines on your behalf. We suggest prompts based on your industry, or you can write your own. Example: "best CRM for startups 2026".
Add competitors
Add the brands you compete with. Lookt tracks them alongside your brand so you can compare visibility, sentiment, and position.
Run your first analysis
Lookt runs your prompts across ChatGPT, Gemini, and Perplexity. The first analysis takes a few minutes. After that, analyses run automatically on a daily schedule.
Key concepts
Workspace
One brand you're tracking. Each workspace has its own prompts, competitors, and analytics. Invite team members to collaborate.
Prompt
A question Lookt asks AI engines regularly. Example: "what are the best project management tools". Grouped by topic for organization.
Prompt run
Each time a prompt is sent to an AI engine and a response is received. One prompt generates multiple runs — one per AI model.
Brand mention
When an AI response contains your brand name or alias. Lookt tracks position, sentiment, and whether AI actively recommended you.
Source citation
Websites AI cites when forming answers. Perplexity and Gemini cite sources most often. Understanding cited sources reveals what content AI trusts.
Overview dashboard
The Overview page answers: "How is my brand doing in AI right now?"
At the top, three key metrics for your brand:
The Visibility over time chart shows daily trends. The Brand comparison table ranks all tracked brands side by side.
The Recent AI responses feed shows the latest queries with source labels (GPT, GM, PX).
Prompts
Prompts are the questions Lookt sends to AI engines. Good prompts mirror how real users ask about your product category.
Writing good prompts
Prompt states
Brands
Manage your brand and competitors.
Your brand
Set during onboarding. Lookt uses pattern matching on AI responses to detect mentions. Add aliases (abbreviations, domain names) to improve accuracy.
Competitors
Track up to 10 competitors alongside your brand. Compare visibility, sentiment, and position per prompt and per model. Lookt also auto-discovers competitors mentioned in AI responses.
Sources
Which websites AI cites when answering questions in your space. This tells you what content AI trusts.
Two views
Gap analysis
Toggle gap analysis to see sources where competitors are cited but you aren't. If G2 is a top source and your competitor has a profile but you don't — that's a gap.
Actions
AI-generated recommendations based on your data. They answer: "What should I do to improve my AI visibility?"
Each action includes:
Actions are categorized by type: on-page, off-page, and technical. Click "Generate" to create new actions. They're deduplicated automatically.
Lookt AI
A chat interface for asking questions about your brand data in natural language.
What you can ask
Query limits
Free plans include 5 queries per month. Paid plans include more. Remaining queries shown at the top of the chat.
Team management
Invite team members from Settings > People. Three roles:
Billing & plans
Manage your subscription from Settings > Billing. View current plan, usage, and upgrade or downgrade anytime. Payments processed through Stripe.
API keys
Generate API keys from Settings > API Keys. All endpoints require authentication via the X-API-Key header.
curl -H "X-API-Key: your_key_here" \
https://api.trylookt.com/api/v1/dashboard/consolidated?workspace_id=YOUR_ID
Metrics explained
Visibility (%)
Percentage of AI responses that mention your brand. Calculated as: (responses mentioning you / total responses) × 100. A visibility of 60% means 6 out of 10 AI responses name you.
Sentiment (0–100)
How positively AI talks about you. 50 is neutral, 70+ is strong positive. Calculated by analyzing the context around each brand mention.
Average Position
Your average rank when mentioned. #1 means AI names you first. If AI lists "1. Competitor A, 2. Your Brand", your position is 2.
Recommendation rate
Percentage of mentions where AI actively recommends your brand — not just lists it. Being mentioned is good. Being recommended is better.
Unbiased mode
By default, Lookt includes all prompts — including ones that contain brand names (e.g., "is HubSpot good for startups?"). These naturally bias toward mentioning the named brand.
Unbiased mode filters out brand-specific prompts. This answers: "When someone asks a generic question about my category, does AI mention my brand?"
Toggle from the Overview page. Expect fewer data points — that's by design. The signal is cleaner.
FAQ
How often does Lookt run analyses?
Daily for active prompts. You can also trigger a manual run from the Prompts page.
Which AI models does Lookt track?
ChatGPT, Gemini, and Perplexity. Enable or disable specific models from Settings > Project.
Can I export my data?
Yes. Export brand comparison data and sources to CSV. API access is also available.
What does position #1.2 mean?
It's an average. If you're #1 in some responses and #2 in others, the average is 1.2. Lower is better.
Why are numbers different in unbiased mode?
It excludes brand-specific prompts, so the data pool is smaller. Cleaner organic signal, fewer data points.
Need help? Email support@trylookt.com