Feasibility of using Flex (RIA)

Hello.

Now we are developing one Internet service, and there is the next task.

There is a large database of product descriptions of 30-50 thousand positions (MySQL is not local). The price is loaded into the system and it recognizes each position and compares it with the existing product descriptions. From 40 to 10 percent of unrecognized positions naturally appear and the task is to manually compare these items with product descriptions. And here the problems begin.
')

Well, at first briefly about the procedure: the administrator selects an unrecognized position from the list, then he looks through the product description in the appropriate section, and if necessary, details the search using filters. If the product description is found - binds them, if not, creates a new product description.

At first they wanted a web interface, but:
Minuses : Low speed (no full grids, no hot buttons) and all that. In my opinion, manipulating a large amount of data in a browser is not the best. Roughly speaking, if I need to edit the tables, I’d better open Excel than Google Docs.
Pros : Directly work with the base

then thinking about the Win32 application
Pros : high operator speed
Cons : atsky crap with synchronization with MySQL,

We decided to stay at a lesser evil: web interfaces.

And then I remembered RIA - Flex.

How do you think it is more reasonable (faster, cheaper) to make a Flex application for such purposes. Or a better web interface. The task - with minimal hemorrhoids, to create an acceptable tool for the operator.

Having rummaged, I learned that Flex cannot work directly with MySQL and I have to do some kind of layer between Flex and the database, is this true?

I do not know whether my topic will be in the format of this blog, but I still decided to take a chance.