You ask.
In plain language, in the NetEnds UI — or by Slack direct message when your host can reach Slack.
Your network assistant
Gia
Plain-language answers from live data, by a model on your own host.
14 days · Every tier · On your own host
Your network, in plain language.
Gia is your network assistant. Ask a question: Gia looks it up in live data, shows you the lookups it made, and answers — from a model running on a server you own.
GiaPending actions · Which devices had config changes in the last 24 hours?
Looked up · Config versions, last 24 h Looked up · Who committed, from syslog
9 devices had configuration changes in the last 24 hours.
Most were manual CLI edits by admin, across morning, afternoon and evening windows. One change was made by root on SW05.
| Device | Type | Version | Source | + Lines | − Lines |
|---|---|---|---|---|---|
| FW01 | Running | v78 | Scheduled | +516 | −461 |
| SW04 | Startup | v19 | On change | +11 | −5 |
| SW03 | Startup | v19 | On change | +6 | −8 |
25 rowsExport
Illustration in the product’s light theme, redrawn from a real Gia answer
How it works
Ask. It looks.
It answers.
It looks.
Gia uses the product’s own tools — inventory, alerts, config history, logs — and shows you each lookup it made. It can check a device too: ping, traceroute and read-only show commands, each confirmed by you first.
It answers.
From live data, with the rows behind the answer — ready to export.
Privacy
Runs where
your data lives.
The model server ships with NetEnds and runs on your host. NetEnds only accepts a model endpoint on your own network — and refuses one outside it.
Your NetEnds host
NetEnds UI
Gia
Model server
Hosted model servicesNot in the path. Nothing to sign up for.
No web searchNo cloud fallback modelNothing to sign up for
Ask by Slack DM and the conversation travels through Slack, over a connection your host opens outbound. The model still runs on your host.
Approvals
It works within your permissions.
Gia uses your role and your scope — never more. A change to a device starts as a draft and waits for a person to confirm it.
Knowledge
It reads your runbooks too.
Connect Confluence or SharePoint and Gia answers with your team’s procedures alongside live data.
What it needs
Your hardware,
sized for the model.
A GPU for speed.
Without a GPU, Gia runs on the CPU: on a 16-core host, a multi-step answer takes one to five minutes. On a 48 GB NVIDIA GPU, answers take seconds, and the installer puts Gia on the GPU when Docker can use it.
Memory and disk.
Size the host for the model — the sizing guide says how.
Your choice.
Install NetEnds with Gia or without it. Everything else runs either way.
Ask your network.
Gia is the third tier — and it’s switched on in the fourteen-day trial, on a host you own.