What Topolux is
Topolux is an operations assistant for enterprise campus networks. It connects to your switches and other network devices, runs inspections and checks from templates, records what it finds in an issue list and produces reports.
It has three parts. The cloud platform handles task scheduling, evidence, analysis and reports. The web console and the mobile app are where you work and review results. Executors are the only part that connects to devices, and they run inside your network or on your phone.
The problems it solves
- Inspections mean logging in to devices one by one, and the results end up scattered across people’s laptops. Topolux runs inspections on every device from a template, on a schedule if you like. Reports are kept in one place, and each device’s inspection history is always available.
- Problems get found, nobody follows up, and they come back. Problems found by inspections go into the issue list, where they are tracked by status and can be reopened if they recur. Decisions that need you arrive as cards in your to-do list.
- No one can say what a conclusion is based on. Every finding carries the command output and the time it was collected, so anyone can check the original.
Three principles
Conclusions come from evidence
Every judgment in an inspection or check comes from the raw output collected in that run. Where evidence is missing, Topolux says so instead of guessing.
Only a confirmed baseline counts Rolling out during the trial
The business baseline records what your network looks like when it is healthy, in four layers: structure, state, configuration and metrics. What the system infers becomes part of the baseline only after you confirm it. Anomalies are judged only against the confirmed baseline, so a long-standing fault is never mistaken for normal.
Execution is controlled
- Topolux runs only read-only commands and limited connectivity probes (ping and tracert), and currently never changes device configuration.
- The platform verifies which devices each task may reach and which commands it may run, then issues the task to an executor. The executor does exactly what was issued and never starts anything on its own.
- Analysis that uses a large language model produces proposals only. The model never connects to devices and never approves anything for you. Raw configurations and logs are never sent to it, and whatever does reach it is redacted first.
How a task runs
Everything that produces a result is a task. Inspections, connectivity checks and diagnoses all follow the same flow:
- Start: on a schedule, or when you start it yourself.
- Collect: an authorized executor connects to the devices, runs read-only commands and sends the raw output back to the platform.
- Analyze: each item is checked against thresholds and rules, and anomalies and gaps in evidence are flagged.
- Report: an inspection report is generated and can be downloaded as a PDF. Findings go into the issue list.
- Follow up: decisions that need you arrive as cards in your to-do list. One example is whether to keep trusting a device whose host key has changed.
Every step and every piece of evidence stays on the task page. You can retry or cancel a task and give feedback on its results.
Where it fits
- Networks: enterprise campus networks.
- Devices: Huawei (VRP 5) and H3C (Comware 7) network devices, collected with read-only commands over SSH or Telnet. SNMP collection requires a standalone executor.
- Availability: currently by invitation.
See Features for details.