Quickstart
Connect a source and get your first investigated answer — in about five minutes.
This walkthrough uses PostgreSQL, but every connector follows the same shape: connect → map → investigate → keep.
Before you start
You need a team and the Member role or higher. A team admin can invite you from Settings → Members.
Open the data catalogue
From the sidebar, choose Data catalogue. The Overview tab shows how many sources you have against your plan limit, how many datasets exist, and anything that needs attention.
Add a source
Click Add data, then pick PostgreSQL from Quick add. Choose where the query should run:
- Cloud — Chartizer connects directly. Fastest to set up; your database must accept connections from our IP range.
- Local agent — a small process you run on your own machine or network. Credentials and query execution never leave your infrastructure.
Fill in the connection details and click Create connection.
psql "postgresql://readonly_user@db.internal:5432/analytics" -c "select 1"
Check what came through
Chartizer introspects the source and its tables, views, or indices appear on the Data sets tab, each with a type and a sync status. Anything that needs a decision — a new table, an unresolved field — lands in Needs attention on the Overview tab.
Field mapping across sources — which is what lets two systems’ data be joined — is coming soon. Today a chart or investigation works within one dataset.
Ask it to investigate
This is the part worth learning first. Rather than building a chart and reading it yourself, give Chartizer the question:
› Which dealers grew fastest last quarter, and what changed for them?
It plans an approach, then works through your data with its own tools — profiling fields, running aggregate queries, segmenting, checking for anomalies — and comes back with a written finding, the charts that show it, and the queries behind every claim.
Each query is shown to you before it runs, and executes only once you accept it. Nothing touches your data unapproved.
Read the evidence, not just the answer
Every statement expands to show the datasets, the query, the row counts, and any caveat — a partial period, a sync gap, unmatched records. It is what separates an answer you can act on from a plausible-looking one. See Investigations.
Keep what is useful
Any chart from a finding can be saved into a project with Save to project. From then on it is a normal chart, bound to its dataset and refreshing with it — no AI involved. Group several onto a dashboard and share it with your team.
You're set up
The source appears under Sources, its data under Data sets, sync problems in Needs attention, and you have asked your first question.