AI Inside GTSuite

AI where it removes work, not where it demonstrates novelty

Every business platform now claims to have AI in it. The useful question is narrower: which specific piece of tedious work does it take off somebody's desk? What follows is answered in those terms.

Your application in another language, this afternoon

Translating a business application is normally a project. Every caption, column heading, label and message has to be found, given an identifier, handed to a translator and put back, and the whole exercise starts again the next time somebody adds a screen.

In GTSuite it is three operations, and none of them requires a developer. The system collects the texts from the application itself, asks the AI to translate them, and saves the result. The application is then available in that language.

When the application changes later, only what is new gets collected, so keeping the translations current is a small recurring gesture rather than a repeat of the original project.

Technical note: texts are identified by their content rather than by their position in the definitions. A phrase used as a column heading, a form label and a tab title is one text, translated once. This is what makes the volume manageable and the incremental update possible.

The end of retyping what a document already says

Somebody in every company spends part of their day reading a supplier invoice, a utility bill, an emailed order or a scanned form, and typing its contents into a screen. It is slow, it is dull, and it is where errors come from.

GTSuite lets you train an assistant on the documents your company actually receives. You show it examples and correct what it gets wrong; each correction becomes a rule it keeps. Corrections are given in plain language, so the person who does the job is the person who trains it, with no technical skills involved. A document type is usually reliable after a handful of examples.

From then on it is available from the page itself. The operator opens the assistant, pastes or attaches the document, and the rows appear in the grid ready to be checked and saved. Different document types, or different suppliers with different layouts, are simply different trained configurations chosen from a list.

What matters here is that the assistant knows the destination. It is not producing an approximate table for someone to rearrange: it fills the exact fields of that page, in the formats that page expects.

Reports without a report designer

Producing a new printed document normally means opening a report designer and laying it out by hand. Instead you can describe what you want, and show an example of the document you produce today, and get a working template back, complete with layout and formatting.

Templates are kept with their version history, so a change can be reversed. And the same templates are not confined to printing: the same technology can place a formatted panel inside a page, showing the selected record as a card or a summary rather than a row in a table.

Analysis where the data already is

The most natural use of AI in a business system is the one that needs no new interface: you are looking at a record, and you want an opinion about it. A candidate against a role, this period against the last, a document against your criteria.

In GTSuite the analysis runs from the page you are already on, with the data of the record in front of you, and answers in the form the question deserves: an assessment, a comparison, a ranking with the reasoning attached. It is a conversation, so you can push back and ask again.

And in building the application itself

There is a further use, less visible from outside, and worth stating plainly because it is unusual. Because a GTSuite application is a structured description rather than program code, AI can assist in producing that description: it can help build the application, not merely run inside it once built.

We use this ourselves, within a controlled process in which nothing reaches a live application without passing validation and human review. It is a large part of why new work moves as quickly as it does.

Try the parts you care about

All of this is easier to judge than to read about. Confirm an email address and you get a live system, the technical documentation, and exercises that show how it is put together.