AI Builder & Large Uploads

The AI Builder

The onboarding wizard designs and provisions an application from a description. The AI Builder is what keeps that loop open afterwards.

Where to find it

Architect Panel → Data:

• AI Builder — describe a change and review what it proposes

The problem it solves

Once v1 exists, the AI that built it traditionally stops being available, and every subsequent change means learning the platform. "Add a Suppliers table", "make Phone required", "we do not use Fax any more" are all things you should be able to just say.

How to ask

Describe the outcome rather than the mechanism. "Track which supplier each part comes from" gives it enough to design the relationship; "add a foreign key" presumes an implementation that may not be the right one.

Review before applying

Changes are proposed for you to look at. Read them, particularly anything touching existing data — adding a field is safe, changing a type or making something required interacts with rows that already exist.

It works within the platform

What it produces is ordinary platform configuration: datastores, fields, forms, permissions. There is no separate AI-built layer to maintain, and anything it creates can be edited by hand afterwards.

Where to be careful

Structural changes to a live system deserve the same care as any other. Try them on a development instance first where you have one, and avoid making sweeping changes in the middle of a working day.

Uploading Spreadsheets and Large Files

Describing a data model in prose is hard work when you already have it in a spreadsheet. You can upload the file instead.

Where to find it

Architect Panel → Data:

• AI Builder — upload a file alongside your description

Working from a spreadsheet

Upload the workbook and the builder reads its structure — columns, types, the values that appear — and proposes datastores and fields to match. That is usually far closer to what you want than a description, because the spreadsheet already encodes decisions you would otherwise have to remember to mention.

Tidy it first

One header row, one table per sheet, no merged cells, no notes rows above the headers. A spreadsheet organised for a human reader gives a worse result than a plain rectangle of data, because the structure it infers is the structure you gave it.

Large files

Uploads are chunked, so a file can be roughly a gigabyte rather than the 64 MB a single request allows. The browser splits the file, the pieces are reassembled on the server, and — because assembling a large file is not a web request — the work happens as a background job.

The tasks involved

• Background Jobs — dispatches and reaps queued work including file assembly. Every minute.
• Document Text Extraction — indexes the contents so uploaded documents are searchable. Every fifteen minutes.

Both ship disabled under Automation → Tasks. A large upload that appears to stall at the end is usually the jobs task not running.

What not to upload

The file contents reach the AI provider. Do not upload live personal data to design a schema — a sample with realistic structure and invented values gives the same result without the disclosure.