Skip to main content

​
Overview

This guide walks you through creating your first workspace, adding prompts, scheduling automated runs, and viewing analysis results.
This quickstart assumes you have access to a running OneGlanse instance. For local development setup, see Local Setup.

​
Step 1: Create your workspace

Workspaces organize monitoring by brand. Each workspace has its own prompts, schedule, and team members.
1

Sign up or log in

Navigate to your OneGlanse instance and authenticate using Google OAuth (configured via Better Auth).
2

Create a new workspace

Click Create Workspace and provide:
  • Name: Your brand or product name (e.g., “Acme Inc”)
  • Domain: Your primary website domain (e.g., “acme.com”)
  • Slug: URL-friendly identifier (e.g., “acme-inc”)
The domain is used in brand analysis to identify your brand in AI responses.
3

Configure providers

Select which AI providers to monitor:
  • ChatGPT (OpenAI)
  • Claude (Anthropic)
  • Perplexity
  • Gemini (Google)
  • AI Overview (Google Search)
Start with 2-3 providers to get faster results, then expand.

​
Step 2: Add prompts

Prompts represent how users might ask about your product category. These are the queries OneGlanse will test across AI providers.
1

Navigate to Prompts

In your workspace, go to the Prompts page.
2

Click Add Prompt

Enter a prompt that represents a common user query in your category.Good prompt examples:
  • “What are the best project management tools for remote teams?”
  • “Which CRM software should I use for a small business?”
  • “What’s the most reliable email marketing platform?”
Avoid overly specific prompts:
  • ❌ “Tell me about Acme Inc” (too direct)
  • ❌ “What does acme.com do?” (too branded)
3

Add multiple prompts

Click Add Prompt to create 3-5 prompts covering different angles:
  • Comparison queries (“best tools for X”)
  • Problem-solving queries (“how to solve Y”)
  • Feature queries (“tools with Z feature”)
Each prompt runs across all enabled providers on every scheduled execution. More prompts = longer execution time.
4

Save your prompts

Click Save Changes to persist your prompts to the database.

​
Step 3: Schedule automated runs

OneGlanse uses cron schedules to automate prompt execution across all providers.
1

Navigate to Schedule

Go to the Schedule page in your workspace.
2

Choose a frequency

Select how often prompts should run:
  • Every 12 hours: Twice daily (good for active monitoring)
  • Every day at midnight: Once daily (balanced approach)
  • Every 2 days at midnight: Less frequent (cost-effective)
  • Every week (Sunday midnight): Weekly snapshots
Times are displayed in your local timezone but stored as UTC in the database.
3

Save schedule

Click Save Schedule. OneGlanse will:
  1. Run prompts immediately for the first execution
  2. Queue jobs for each provider in Redis/BullMQ
  3. Agent workers will consume jobs and execute prompts in real browsers
  4. Follow the cron schedule for future runs
4

Monitor execution

Check the Last Prompt Run timestamp to confirm execution started.
Initial runs take 2-5 minutes per provider depending on response complexity and browser automation timing.

​
Step 4: View results

Once prompts execute and responses are analyzed, view insights in the Dashboard and Prompts pages.
1

Navigate to Dashboard

Go to the Dashboard page to see aggregated metrics:
  • Presence Rate: Percentage of responses mentioning your brand
  • Average Rank: Your position when AI lists alternatives
  • Top Source: Most-cited source domain for your brand
  • Top Competitor: Most frequently mentioned competitor
Use model and time filters to narrow results.
2

View prompt details

Go to the Prompts page to see performance by prompt:
  • GEO Score: Overall brand performance (0-100)
  • Sentiment: How favorably your brand is portrayed
  • Visibility: Prominence of your brand mention (0-100%)
  • Position: Rank in lists (1 = top position)
Click a prompt row to expand and see full responses per provider.
3

Analyze responses

In the prompt detail view:
  • Read the full AI response
  • See extracted metrics (GEO Score, sentiment, visibility, position)
  • View cited sources with links
  • Check if analysis is complete (“Analysis in progress…” if pending)
4

Export data

Use the Export button to download:
  • JSON: Full analysis data with metadata and recommendations
  • CSV: Tabular data for Excel/Sheets analysis

​
Understanding your first results

​
If your brand is mentioned

Look for:
  • High GEO Score (60+): Your brand has strong visibility
  • Top 3 position: You’re prominently recommended
  • Positive sentiment (70+): AI portrays you favorably
  • High visibility (70%+): Your mention is prominent in the response

​
If your brand is NOT mentioned

This is common and expected. It means:
  • AI providers don’t yet associate your brand with this query
  • Your content may not be indexed or prioritized by AI training data
  • Competitors have stronger signals for this query
Next steps:
  • Refine prompts to be more specific to your niche
  • Create content that targets these queries (GEO optimization)
  • Track competitors to understand what they’re doing differently
Brand analysis defaults to zero values when your brand is not mentioned. This is intentional and not an error.

​
What’s next?

Understand workspaces

Learn about workspace organization, team members, and role management.

Prompt best practices

Write effective prompts that surface meaningful brand mentions.

Interpret metrics

Deep dive into GEO Score, sentiment, visibility, and position metrics.

Team collaboration

Invite team members and manage workspace permissions.

​
Troubleshooting

Check:
  • Schedule is enabled in Schedule page
  • Last Prompt Run timestamp updates after schedule time
  • Agent workers are running (pnpm dev:agent in development)
  • Redis is accessible and BullMQ queues are healthy
Analysis runs asynchronously after responses are captured. If stuck:
  • Wait 2-5 minutes for OpenAI analysis to complete
  • Check that OPENAI_API_KEY is configured in environment
  • Review agent worker logs for analysis job errors
This is normal if:
  • Your brand is new or has low visibility
  • Prompts are too broad or not relevant to your niche
  • AI training data doesn’t include strong signals for your brand
Try:
  • More specific prompts targeting your exact use case
  • Including your brand’s primary keywords in prompts
  • Testing different provider combinations
Browser automation can fail due to:
  • Provider UI changes (prompts may need updating)
  • Rate limiting or bot detection
  • Network timeouts
Check agent worker logs for Playwright errors.

​
Need help?

Join the community

Report issues, request features, or contribute to OneGlanse on GitHub.

Build docs developers (and LLMs) love

Get started for freeTalk to us