Performance and method in complex queries

Very good morning to all.

I am testing your technology and in addition to congratulating you on it, I would like to ask you some questions:

Example

The answer is:

-- CREATE TABLE [dbo].[clients](
-- 	[id] [int] IDENTITY(1,1) NOT NULL,
-- 	[name] [nvarchar](10) NOT NULL,
-- 	[city] [nvarchar](100) NOT NULL,
-- 	[country] [nvarchar](100) NOT NULL
-- ) ON [PRIMARY]
show me clients on Alejandro Korn

The response is:

SELECT * FROM clients WHERE name = ‘Alejandro Korn’

However, if the input is:

-- CREATE TABLE [dbo].[clients]
-- (
--     [id] [int] IDENTITY(1,1) NOT NULL,
--     [name] [nchar](10) NOT NULL,
--     [city] [nchar](100) NOT NULL,
--     [country] [nchar](100) NOT NULL
-- ) ON [PRIMARY]
-- INSERT [dbo].[clients] ([id], [name], [city], [country]) VALUES (2, N'Client One', N'San Francisco', N'USA')
-- INSERT [dbo].[clients] ([id], [name], [city], [country]) VALUES (3, N'Client Two', Alejandro Korn', N'Argentina')
show me clients on alejandro korn

The response is:

SELECT * FROM clients WHERE city = ‘Alejandro Korn’

The question is: Is there an EFFICIENT way to train this model with the corresponding values ​​and not affect the current limitation of 4000 tokens for more complex models with more data?

Even using only the database structure as input in complex models would limit the number of tokens.

How could I approach this?

From already thank you very much. Greetings