Kicking off the AgilePoint version 10 series of blogs, we’re back with enhancements to a capability that has wowed customers, partners and analysts alike – using Generative AI to create complete applications that include processes, forms and data tables.
Wait a minute, haven’t I seen this before?
Yes, well, kind of.
If you follow our product blogs, you might have seen a similar capability previously, where Generative AI used a prompt to generate the complete application. While cool and in-line with what other major vendors are doing, there are some drawbacks to this approach. Namely that the application generated never truly captured each customer’s unique essence. In short, it was pretty generic and as a result, not really viable for Enterprise use.
Ok then, what’s changed?
Using a prompt is still an option and won’t be going away any time soon, but as of now there are much better ways to generate applications via AI. Most importantly, ways that can reduce hallucinations and ensure the resulting application is hyper-relevant to your organization, and exactly what you’re looking for.
To do so, we need to add Computer Vision to perceive the users intent, and what better way to do that than in the form of something prescriptive like a source document or mockup of the process map and associated interfaces.
How Does This Work?
- Collect a Process Map as an Image or PDF: Draw it in Visio or Figma, export it from a legacy system or download a map that you’ve found online, and simply feed it to AgilePoint GenAI.
- A fine-tuned Computer Vision model will examine the process including swim lanes, process logic and human vs system activities, and intelligent transpose these into AgilePoint artifacts. Feel free to roll with the suggestions, or make changes until everything is just right.
- Find a Form or Interface your Process will use: Again, this can be exported from another system, designed in your favorite mock up tools, or sketch it out on a piece of paper. (Yes, the sketching approach actually works)
- Similar to the process, Vision Model will understand the form, and again, bring those over into AgilePoint UI elements, and provides the designer with the opportunity to make changes as needed.
- Finally, based on the fields for the form AgilePoint will also generate tables in Data Entity, where you’ll set up the primary key, or primary and foreign keys in the event of a Master-Detail form.
That’s it! Just those 5 simple steps are required, but optionally once your app has been generated, feel free to add in rules, JavaScript or CSS, or anything else for that matter, and you’re ready to go.
Ready to see it for yourself? Watch an overview below.
Please note that this video is recorded while this feature is in limited-preview beta version. Some of the labels and icons might change by the time it goes GA but the general concept of generating an app would remain the same.