Aviz Network Copilot 1.0 details SNMP and ONES telemetry analysis
Aviz Network Copilot 1.0 is described as an AI-driven network analysis assistant that uses natural-language chat to ingest multi-vendor network telemetry, analyze performance and compliance, and present analytics to operators. For enterprise IT and security teams, the update centers on how network data is collected, processed, secured, and used for monitoring and troubleshooting workflows.
Research Overview
The blog frames AI in network management as the use of machine learning, deep learning, and natural language processing to improve network effectiveness, dependability, and security. It positions Network Copilot 1.0 as an AI-driven tool for network operators to identify performance bottlenecks and resource utilization challenges.
Operators interact with the tool using natural language prompts to obtain insights into network performance metrics and address issues. The approach is described as supporting more efficient monitoring and troubleshooting to maintain network performance and reliability.
Key Capabilities
Data ingestion and storage
The blog states that Network Copilot 1.0 gathers and stores data collected from SNMP and ONES Collector. It specifies that metric collection includes EOS, SONiC, Cumulus, and NXOS, and that the data is acquired, transformed, and preserved in an organized format for later analysis.
It also states that standardized data collection is intended to make the solution ready for multi-vendor on-prem deployments.
Chat interaction and context handling
The blog describes a chat prompt that initiates a conversation and steers it based on user responses. It adds that chat history can be exported using the prompt and that responses can be streamed as a continuous data stream.
For context, it states that network compliance is defined using user preferences and adjustable threshold values. It further says context is imported and controlled through RAG in chromaDB.
Analytics and visualization
The blog describes graphical outputs using pie charts, bar charts, and timeline graphs to present insights, trends, and patterns. It also states the model can summarize, describe, and analyze gathered data and compute averages, counts, and percentages.
Examples include representing inventory with a pie chart and network utilization over time with a line chart, plus an example for average CPU utilization over past three months.
Security controls
Network Copilot 1.0 is described as using HTTPS certificates and a Secure API to secure customer data. The blog states that secure APIs enable interaction between the large language model and the database for data exchange, task execution, and information retrieval.
It also states that LLMs can operate autonomously using custom-built tools to perform designated functions.
Operational Use Cases
Inventory and accounting
The blog describes capturing network device information including hostname, HWSKU, operating system version, interface details, capacity, and device uptime. It says this supports inventory record maintenance and network asset accounting.
It includes an example labeled as a snapshot where the model responds with device details from inventory.
Capacity planning
For capacity planning, the blog describes forecasting network capacity by comparing available bandwidth with utilized bandwidth. It states this helps operators design infrastructure to support current and future network demands.
It includes an example described as a model projecting overall capacity and utilization based on past data.
Anomaly detection
The blog states Network Copilot 1.0 is trained to identify network failures connected to sudden increases in traffic. It describes traffic spikes that can raise CPU and memory utilization and potentially lead to failures in control planes or links.
It says recognition of these patterns supports mitigation of potential disruptions and proactive issue handling before escalation.
Network compliance
The blog states the model is pre-configured with default compliance thresholds that establish limits for relevant captured metrics. It says compliance checks compare observed values for CPU, memory, bandwidth utilization, and packet drops against predefined limits.
It adds that threshold values are customizable by users.
Overall, the blog describes Aviz Network Copilot 1.0 as a chat-based network analysis tool that ingests SNMP and ONES Collector telemetry, applies analysis for inventory, capacity planning, anomaly detection, and compliance checks, and uses HTTPS and secure APIs for data protection. This “Blog Signals brief” is a fact-based summary of the vendor blog.