Fine-tuned for networks
We host our own LLMs, fine-tuned on real network operations workflows. They speak like an operator and offer executable advice.
Simba is the AI assistant built into Axon. Ask in plain language what's happening on your network. Simba calls real tools against your control plane, grounds every claim in live telemetry, and shows you exactly what it ran. No hallucinated stats. Real insights and solutions.

We host our own LLMs, fine-tuned on real network operations workflows. They speak like an operator and offer executable advice.
Simba doesn't summarize. It runs typed tools against your control plane (site queries, policy lookups, health checks), and the result is what it answers with.
Inference and tool calls run inside the private Axon VPN. Telemetry never leaves the trust boundary to produce an answer. Every tool call is logged and inspectable.
When Simba checks whether your Development site is healthy, it runs find_overloaded_switches(site: "Development") against the control plane. The JSON result is right there in the chat. The answer above it summarizes that JSON and leverages our proprietary models to offer valuable insight.


Simba walks your block-rules and rate-limit rules per site, returns the structured result, and answers with a plain summary. If the answer changes tomorrow because someone applied a new rule, the next query reflects it instantly. No cache, only live calls.
How tool-calls reach the control planeA representative slice (there are more, and we ship new ones regularly). Tools that mutate state (apply a policy, revoke a token) always require explicit operator confirmation; Simba can stage them, but never executes them silently.
“How is the Development site doing?”
“The Internet is slow at the Development site, what is causing it?”
“What traffic is blocked on the network?”
“Which switch is closest to its CPU limit?”