Read what Azi proposes

KQL Basics

Goal: Read and tweak the queries Azi writes for you. You do not need to master KQL.

Warm queries are written in KQL (Kusto Query Language). Azi writes it for you, so most of the time you just describe what you want and approve the result. This page is the small amount of KQL worth knowing so you can read a proposal, sanity-check it, and make a quick edit before you save.

The basic structure

ProductData
| where Timestamp > ago(24h)
| summarize signups = count() by bin(Timestamp, 1h)
| order by Timestamp asc

Read it as: take ProductData, keep the last 24 hours, count rows into hourly buckets, sort oldest to newest. That is a complete warm query.

The operators you'll see most

Operator	Does	Example
where	filter rows	where Campaign == "spring"
summarize	aggregate (count, average, ...)	summarize total = count()
by	group the aggregate	summarize count() by Campaign
project	pick columns	project Timestamp, Campaign
order by	sort	order by Timestamp desc
take	first N rows (quick peek)	take 10

Time and aggregation

• Time windows: ago(1h), ago(24h), ago(7d) - relative to now.
• Bucketing: bin(Timestamp, 1h) groups rows into hourly buckets (use 1d for daily).
• Aggregations: count(), avg(value), sum(value), min(value), max(value).

Reading one of Azi's proposals

When Azi proposes a query, scan it in five steps:

1. Table - what data is it reading? (the first line)
2. where - what is it filtering to? (time range, a campaign, a threshold)
3. summarize - what is it counting or averaging?
4. by - what is it grouping by? (per hour, per campaign)
5. order by - how is it sorted?

If those five match what you asked for, approve it.

Quick edits you can make by hand

Want to change	Edit
The time window	ago(24h) becomes ago(7d)
The threshold	> 40 becomes > 100
The grouping	by bin(Timestamp, 1h) becomes by Campaign
The metric	count() becomes avg(Value)

You can edit the query text directly in the Warm Query Manager, run a preview, then save. Or just ask Azi to make the change - see Working with Azi.

Next steps

• Build queries by describing them -> Working with Azi
• The shapes that come up most -> Query Patterns
• Save one as a shared source of truth -> Warm Queries