Skip to main content

Meet Your Assistant

Every Waxell workspace includes an AI operations advisor. It reads your live data, knows the documentation, and can propose real actions — always with you approving the click. This tutorial walks you through your first session; ten minutes from "what is this button" to a governed fleet with an alert rule drafted.

Prerequisites​

1. Get your fleet briefing​

Click Ask AI in the top navigation — the assistant dock opens on the right. Ask:

"Give me a full fleet briefing"

You'll get your agents, run counts, error rates, and spend — real numbers from your workspace, with a chart when the data warrants one. This is the fastest way to orient yourself any morning.

Try a follow-up with specifics:

"What did I spend on LLMs in the last 24 hours, broken down by model?"

Notice the answers name your agents and your models. The assistant queries the same APIs the dashboards use — it never answers data questions from general knowledge.

2. Ask how something works​

The assistant doubles as the documentation:

"How do I instrument a Python agent with waxell-observe?"

The answer comes from the official docs with cited sources — every claim links to the page it came from. Troubleshooting works the same way; paste an error and it searches the docs with your error text.

3. Create your first policy — conversationally​

"Help me set up policies for my agents"

The assistant reads your fleet first: which agents are active, how much they run, what coverage already exists. Then it proposes policy cards — complete configurations you can review. Each card has Create Policy and Open in Editor buttons.

Ask it to explain before you commit:

"Explain this proposed policy"

Then click Create Policy on the one you want. That click is the approval — the assistant never creates anything without it.

4. Let it propose actions​

Beyond policies, the assistant proposes small concrete next steps as action cards — adding items to your setup plan, drafting an alert rule, prepping a teammate invite. Try:

"What should I set up next?"

Approve a card and the step lands in your plan on the Getting Started page.

5. Check your Feed​

Switch the dock to the Feed tab. This is where the assistant's proactive findings land — setup nudges, new failure patterns, spend changes — each linking to a pre-filled conversation. An unread badge on the tab tells you when something new arrived, and a weekly digest email summarizes the highlights.

Your briefing page at /assistant collects the same findings alongside your recent conversations.

6. Where to go from here​

  • Ask "What can you do?" any time for the full capability tour
  • Use ⌘K and type a question — the assistant is wired into the command palette
  • If it can't resolve something, Talk to a human at the bottom of the dock hands off your transcript to support

What you learned​

CapabilityThe ask
Live data with charts"Give me a fleet briefing"
Cited documentation answers"How do I …?"
Policy creation with approval"Help me set up policies"
Action cards"What should I set up next?"
Proactive findingsThe Feed tab + weekly digest

Next Steps​