We have used PB Open methodology to modernize PB Applications for several years. The risk is mitigated, but the time is still significant. Could we benefit from the evolutions of AI models?
A set of tools leveraging AI models, without permanently linking to any of them, evolving continuously and reusing the prompting system across projects.
Basically all. JS-based frontends are recommended.
Basically all. C# with .NET DataStore is recommended to reduce conversion risk.
ā Recommended for best results and lowest conversion risk
A methodology assisted by tools ā not a packaged product.
We do not sell the tools ā we use them to speed up development, free of charge (except for LLM costs). We provide full resources for each engagement.
A structured iteration from policy to production.
Each iteration follows the same sequence ā policy definition, scope selection, test catalog, conversion, manual completion, and production release.
Definitely.
We can create a hybrid team and define the best way to work together ā adapting the engagement model to your team's skills, availability and preferences.
No. Convert only what needs to run on Web or Mobile.
The full application integrates with the converted module. PowerServer keeps the main application cloud-accessible while migration proceeds at your pace.
Before any full engagement, we run a paid pilot on a real module from your application. You see the output, validate the quality, and decide whether to proceed.
After the initial presales talks, we run a consulting study to define the UI/UX and the coding style. The first target will be selected for the conversion, and an estimation model is provided.
Once the estimation is accepted, the first selection is converted, compliant with the defined style. These windows can run integrated with your own application, and also directly from the browser. At delivery, the results are evaluated again, the styles refined, and one or two additional iterations might be necessary to finalize the result and the styling decisions.
For every successive iteration: a number of windows (or a full workflow or module) is selected, the estimation model is used for providing estimations, and upon approval the selected subset is converted, tested and delivered.