Project Development Assistant

Inference engine and smart reference through semantic search on sentence embeddings.

CONCEPT  User opens application. User says "I want to start a new engineering project on Sensor Fusion for Altitude Detection". The application allocates a distinct context space for this new project. The user can input information about a project. This information may be anything relevant like a feature, a requirement, a change, a review, etc. The application saves this information in the respective project memory space. For each new input, the application reviews the existing project memory and retrieves:

1. Relevant previous entered project information, and duplications.
2. Contradictions or conflicts between the new input and the project memory.
3. Groupings identified in the project memory, to classify the new input.
4. Relevant Inferences made from the project memory that may not have been explicitly stated.

This would be a local desktop application. User-entered text would be saved as vector / sentence embeddings. Similarity measures (semantic search, and cross-encoder) would be used to generate the outputs. Some kind of interface would have to be made to make this output interpretable and productive.