API Maker
The backend engine for the AI era
ER diagramEpisode 5 4:56 9 chapters November 2026
A few keys in a schema link two tables, whatever their engines: the shipments of the warehouse meet the orders of the shop. Typos are caught when the schema is saved, and deep populate, find and join, master save and the ER diagram all follow the relation.
In the playlists The complete course Databases & schemas
Tables are often linked to each other. In API Maker, a few lines in a schema link two tables, even when they are in different databases. Deep populate, find and join, master save, and the ER diagram all follow these links. In this video, the shipments of the warehouse meet the orders of the shop.
When data is spread over many databases, things get hard. You write joins in your application, call by call. Foreign keys stop at the edge of one database. And someone draws a diagram by hand, which is out of date a week later. With API Maker, a relation is a few keys in a schema, between any two databases. The server checks every name when you save. And the ER diagram is drawn from the schemas.
In this video: a relation typed in a schema, two typing mistakes caught, the relation at work, the ER diagram of six databases, and one table with its links.
Open Instance API. Pick sqlserver, then warehouse, then dbo.shipments, and open its schema. A shipment in SQL Server ships an order from MySQL. So under order_id, we type the instance mysql, the database, the table orders, and its column id. We make two typing mistakes, the way they happen in real life. Click Update. It is refused. order_id names a column, but no table, because tabel is not a valid key.
We fix it to table. Click Update again. It is saved, with a warning: there is no schema for mysql, sale, orders. The database name is wrong. We fix it to sales. Click Update and Close. Now it is saved, with nothing to warn about.
So a relation is a few keys on a column. When the table is somewhere else, you give the instance and the database. Then you give the table and its column. Any column works: an ID, a SKU, or a coupon code. Inside one SQL database, the foreign keys became relations by themselves, when the schema was made.
Now let us see it work. Open API Testing, pick sqlserver, then warehouse, then dbo.shipments, and the API Get all. Ask for shipment one, and add deep on order_id. Send. We get the shipment from SQL Server, and inside it, its order from MySQL. One call, two databases.
Now open Database, then ER Diagram. Tick the six instances: MongoDB, MySQL, MariaDB, PostgreSQL, SQL Server, and Oracle. Click Apply Selection. You see every table of the account, and every relation, across the databases. Hide the list for more room, click Auto-Arrange, and open the legend. It shows primary keys, foreign keys, and required fields. Now click the search box, and type shipments. Pick the table, and its relations light up.
warehouse_id goes to the warehouses, in the same database. And there is order_id. Fit the view again, and the line of order_id reaches the orders, in MySQL.
Good to know. Master save, deep populate, and find and join follow these relations. Videos nine, ten, and twelve show them. Virtual fields describe the other side, one to many. Video eleven shows it. The diagram can be saved as PNG or SVG. And a relation is part of the schema, so it goes to Git with it.
The links between tables are in one place: the schema. Every API and the diagram read it. Every name is checked when you save. A key that does not exist is refused, and a table that is not there is reported. So a relation written by an AI assistant gets the same checks.
To sum up. A relation is a few keys in a schema, across databases. Its names are checked when you save. Deep populate and the other APIs follow it. And the ER diagram shows the whole account. Next: a tour of the generated APIs. API Maker. Build faster, spend smarter, deliver better.
Length 5:25Generated APIs
Length 10:05Six engines
Length 6:49Databases
Length 12:35Table designer