Module 2 · Lecture 2.3
Veezoo as Co-Builder
Work on your own branch and let the AI co-builder change the Knowledge Graph from chat, one confirmed proposal at a time.
The worked example now has a freshly generated Knowledge Graph. So how do you change one? The first habit comes first: never work on Main, where your Viewers ask their questions. Create your own branch and work there.
Then the lecture introduces your new colleague. The same chat that answers questions can also change the Knowledge Graph. You describe the job in plain English, Veezoo works through the files, and every change comes back to you as a proposal to confirm.
Key points
- Branch first: Viewers live on Main, so Creators work on their own branch and push to Main when done.
- The Optimize Knowledge Graph button runs a best-practice pass, and asking in chat does the same.
- Agentic modelling: give the co-builder a whole job, such as adding descriptions and synonyms and hiding technical fields.
- Every change comes back as a proposal in chat that you confirm before it lands.
- Auto-apply in the creator tools skips that confirmation, on your own branch only.
- SQL access lets Veezoo inspect real data, such as column values and table relationships, for better modelling decisions.
- One pass leaves no stray ID fields, proper descriptions, and more synonyms.
Check your understanding
You want the co-builder to clean up your new Knowledge Graph. What should you do before asking it to change anything?
The habit is to leave Main to your users. Create your own branch, work there, and push to Main once the changes are ready.
A Creator turns on auto-apply for Knowledge Graph changes in the creator tools. What is true?
Auto-apply is limited to your own branch and suits repetitive small changes. Main always requires explicit confirmation.
What does the SQL access creator tool allow?
Admins grant SQL access to Creators so Veezoo can check real data while modelling. User-facing answers still go through the deterministic translation to SQL.