F12
SNAP – low-code AI engine
SNAP is F12's low-code front-end framework. Solvingo built the AI layer on top of it - the engine that automatically generates, interprets and analyses the framework and its low-code elements, from platform-specific model research through training data and training scripts to evaluation.
Tech stack
Markdown
JSON / XMLThe challenge
The low-code framework has its own element and logic language. For an AI to genuinely generate, interpret and analyse it, you need a platform-specific model - and there is no ready-made public training data or training pipeline for that.
What we delivered
- Research and selection of platform-specific solutions
- Assembling training data from the low-code elements
- Building training scripts and training the model
- Generation, interpretation and analysis over the low-code framework
- Assistive functions based on the low-code elements
- Evaluation and fine-tuning
The solution
Generation & interpretation engine
The model generates and interprets the low-code framework and its elements: a working low-code structure from a description, and readable meaning from the structure.
Platform-specific training
Selecting the right approach, assembling training data from the low-code elements, custom training scripts and training the model.
Evaluation & assistive functions
Evaluating the trained model, then AI assistive functions built on the low-code elements that speed up development.
Measurable results
capabilities: generate, interpret, analyse
platform-specific model and training pipeline
assistive functions built from the elements
A platform-specific AI with measurable capabilities
An engine that generates, interprets and analyses the low-code framework - on bespoke training data and scripts, with evaluation.