Cisco describes AI parsing of offline CLI logs for inventory and troubleshooting
An AI-driven workflow parses offline Cisco CLI outputs into structured inventory data, then correlates that data with lifecycle and bug advisory sources to support interactive troubleshooting and root cause analysis. For enterprise IT and security teams, this addresses the time required to extract software, hardware, and risk context from exported device text files.
Research Overview
The workflow targets exported command outputs such as show version, show platform, and show inventory from Cisco routers and switches, including ASR1001, NCS platforms, ASR9K, and Nexus switches. The source text can include software versions, hardware module details, serial numbers, firmware levels, uptime, and configuration register values that engineers otherwise interpret manually.
The approach is described as two stages: first extracting software and hardware metadata from raw CLI text, then integrating the extracted results with support knowledge used during troubleshooting, including lifecycle and bug advisories. The second stage uses an AI assistant with retrieval-augmented retrieval over both logs and knowledge sources.
Key Findings
The parsing stage is reported to extract fields such as OS version, uptime, image and boot location, and configuration register values from examples of ASR1001 outputs. It also extracts chassis and module information from show inventory, including chassis model and serial numbers, route processor and embedded services processor modules, and transceivers with their serial numbers.
From show platform outputs, the workflow is described as extracting firmware versions, ROMMON information, and CPLD versions for modules. The extracted structured records also include attributes such as physical memory, interface counts, and boot settings, with logic described as applying across Cisco families including NX-OS and IOS XR platforms.
Technical Breakdown
The text-to-structure step converts CLI content into structured fields including device model, OS version, uptime, serial numbers, module inventory, transceiver details, firmware versions, and main memory. The brief states that the extraction logic generalizes across Cisco device families, including NX-OS and IOS XR platforms.
For support correlation and assistance, uploaded logs are broken into chunks and converted into vector embeddings. The system then enables semantic search across logs and knowledge sources, and when an engineer asks a question, it retrieves relevant log fragments and knowledge base entries before generating an answer.
Operational Impact
The brief describes lifecycle and end-of-support correlation by using lifecycle bulletins, including an example that the Cisco ASR1001 platform reached end of support in April 2021. Without an integrated lifecycle dataset, the system is described as unable to answer end-of-life status questions.
It also describes bug and advisory correlation by mapping the detected IOS XE version to known caveats and vulnerabilities, using Cisco release notes and advisory documents. For example, the brief cites bug IDs as part of the workflow, and it states that the assistant can correlate multiple data points to accelerate diagnosis, such as when interface flapping triggers searches for error messages, configuration changes, or known hardware issues.
Blog Signals brief is a fact-based summary of the vendor blog.