- Overview
- Prerequisites
- Initial Setup
- OpenAI API Configuration
- Function Configuration
- Testing Configuration
Made4intelligence is an AI assistant integrated into the made4graph system that offers:
- Intelligent Analysis of cpe: Automatic diagnosis of equipment tr-069
- Service Troubleshooting: Connectivity issues analysis
- Interactive Chat: Conversational assistant for technical support
- Dynamic Configuration: Customizable prompts and templates
- Access OpenAI Platform
- Log in to your account
- Go to "API Keys" in the side menu
- Click on "Create new secret key"
- Copy the generated key (format:
sk-...)
- Log in to the system made4graph
- In the side menu, locate the option "Made4intelligence Config"
- Click to access the settings screen
Note: If the "Made4intelligence Config" option is not visible in the menu, access the user group permissions and enable this feature so that it becomes available.
- On the settings screen, click on "Set up OpenAI"
- Fill in the requested fields
- OpenAI API Key: Paste the copied key
- Model: Select the desired model
- GPT-5 Nano: Compact and fast version
- GPT-5 Mini: Optimized version
- GPT-5: Faster and smarter
- GPT-4o: Faster and smarter ⭐
- GPT-4o-mini: Optimized version
- GPT-4 Turbo: High accuracy
- GPT-4: Full model
- GPT-3.5 Turbo: Good cost-benefit ratio
- What is: Token limit (words/parts of words) that the assistant can use in a response
- Standard: 4000 tokens
- How it works:
- 1 token ≈ 0.75 words in Portuguese
- 4000 tokens ≈ 3000 words
- The more tokens, the longer and more detailed the response
- Recommendations:
- 1000-2000: Short and direct answers
- 3000-4000: Detailed answers (recommended)
- 5000-8000: Very detailed analysis
- Cost impact: More tokens = higher cost per response
- Automatic adjustment: For GPT-5 models, a minimum of 7000 tokens is automatically applied
- What is: Controls the "creativity" and variability of responses
- Scale: 0.0 to 2.0
- Standard: 0.7
- How it works:
- 0.0-0.3: Very consistent and predictable responses
- 0.4-0.7: Balanced between accuracy and creativity ⭐
- 0.8-1.2: More creative and varied
- 1.3-2.0: Very creative, can be less precise
- Recommendations for use:
- 0.1-0.3: Accurate technical analysis
- 0.4-0.7: Diagnostics and troubleshooting (recommended)
- 0.8-1.0: Educational explanations
- 1.1-1.5: Creativity and brainstorming
- Practical example:
- Temperature 0.1: "The equipment has a temperature of 75°C, above the limit of 70°C."
- Temperature 0.7: "The equipment is operating at 75°C, which is 5°C above the critical limit of 70°C. This may indicate..."
- Temperature 1.2: "It seems like your equipment is 'heating up' more than usual! The temperature of 75°C is..."
- Click on "Test Configuration"
- Wait for the system's response
- If everything is correct, click on "Save Settings"
- Critical Temperature: Above which temperature to consider critical (default: 70°C)
- Minimum Uptime: Minimum operating time (default: 24h)
- pon Minimum RX: Minimum reception signal (default: -25 dBm)
- pon Maximum TX: Maximum transmission signal (default: 50 dBmV)
- ✅ Check DNS: Validate DNS configuration
- ✅ Validate ip WAN IPv4: Check address ip
- ✅ Check WiFi status: Analyze WiFi networks
- Empty Data: How to handle values "::" (recommended: Alert)
{{TEMP_CRITICAL}}: Critical temperature set{{UPTIME_MIN}}: Minimum uptime in hours{{PON_RX_MIN}}: pon Minimum RX power in dBm{{PON_TX_MAX}}: pon Maximum TX power in dBmV{{CHECK_DNS}}: Check DNS (true/false){{CHECK_IPV4}}: Check IPv4 (true/false){{CHECK_WIFI}}: Check WiFi (true/false)
- template Basic: Simple and straightforward analysis
- template Detailed: Complete analysis with explanations
- template Technician: Advanced analysis for technicians
To analyze the cpe, we can have a prompt for that and use the available variables:
- Dynamic variables such as
{{SERVICE_NAME}}
- Define the basic behavior of the assistant
- Controls limitations and scope of responses
- Available templates:
- Specialized Technician: Focus on networks and tr-069
- Friendly and Accessible: Simple language
- Restricted and Focused: Only specific questions
- Advanced Diagnosis: Focus on troubleshooting
- Educational and Explanatory: Teaches concepts
The "Service Issue" function allows diagnosing specific connectivity problems related to services such as Netflix, YouTube, WhatsApp, etc.
- User reports an issue with a specific service
- The system analyzes technical data from cpe (DNS, ip, pon, WiFi)
- Correlate user issues with technical data
- Provides diagnosis and specific solutions for the service
- Click on "Add Service"
- Fill in the required fields:
- Service Name: Unique identifier (e.g. netflix, youtube, whatsapp)
- Icon: Select appropriate icon from the extensive list
- Category: streaming, social, gaming, productivity, others
- Description: Brief description of the service
- template of Prompt: Custom prompt for specific analysis
🎬 Streaming:
- Netflix, YouTube, Spotify, Twitch, Amazon Prime
- Disney+, HBO Max, Apple TV+, Paramount+
📱 Social:
- Instagram, Facebook, WhatsApp, Telegram, Discord
- Twitter/X, TikTok, LinkedIn, Snapchat
🎮 Gaming:
- Steam, Xbox, PlayStation, Epic Games, Origin
- Battle.net, Uplay, GOG, Roblox
💼 Productivity:
- Google Drive, Dropbox, Slack, Trello, Notion
- Microsoft Teams, Zoom, Skype, Asana
🛒 Others:
- PayPal, Uber, Airbnb, iFood, Rappi
- Google Maps, Waze, Spotify, Deezer
Each service can have its own template prompt that includes:
- Specific issues with the service
- Relevant technical correlations
- Customized solutions for the service
Scenario: User reports "Netflix is not loading"
System Process:
- Problem analysis: Netflix with loading issue
- Technical check: DNS, WAN ip, pon, WiFi
- Correlation: "DNS issue affecting Netflix"
- Diagnosis: "Misconfigured DNS preventing access to Netflix"
- Solution: "Set up DNS 8.8.8.8 and 8.8.4.4 on the router"
Other Examples:
- WhatsApp: "Can't send messages" → Check connectivity and DNS
- YouTube: "Videos freezing" → Analyzes speed and quality of the connection
- Steam: "Can't download games" → Check bandwidth and firewall
- Accurate diagnosis for specific problems
- Targeted solutions for each service
- Technical correlation between problem and data from cpe
- Personalized experience for the user
- Set up the OpenAI API
- Save settings
- Go back to screen cpe
- Click on the "AI Assistant" button
- The assistant should open normally
- In the wizard, select "Intelligent Diagnosis"
- Wait for the equipment analysis
- Check if the answer is appropriate
- Choose "Check Services"
- Select a configured service
- Test specific analysis
Possible causes:
- Invalid or expired API key
- No credits in the OpenAI account
- Model not available
- Connectivity issue
Solution:
- Check the key on the OpenAI Platform
- Confirm if you have credits available
- Test with another model
Solution:
- Reduce the number of maximum tokens
- Use more economical model (GPT-3.5)
- Simplify the prompts
- Add credits to your OpenAI account
🎉 Setup completed! Made4intelligence is ready to use.