Skip to content

Home

Bro is a local-llm application built around Gemma 3 4B IT in GGUF format. It combines local text generation, Gemma multimodal Image-to-Text, document-grounded retrieval, semantic search, prompt engineering, and SQLite-backed data management in a single application.

Bro

Application capabilities

Capability What Bro provides
Text generation Chat, analysis, reasoning, coding, writing, editing, summarization, extraction, classification, translation, comparison, and structured output.
Image understanding Visible-text extraction, image description, image Q&A, screenshot/chart/diagram analysis, structured image extraction, and image comparison.
Document Q&A Local PDF/TXT/DOCX retrieval with grounding controls, sqlite-vec/cosine search, document actions, and optional Gemma vision OCR.
Semantic search Local chunk embeddings, similarity search, document-aware results, and reusable context routing.
Prompt engineering Category-aware SQLite prompt management using ID, Caption, Name, Category, and Text.
Data management Excel import, SQLite CRUD, filtering, aggregation, visualization, schema administration, indexing, governance metadata, and guarded SQL.

Model modality boundary

Gemma 3 4B IT supports:

  • text input;
  • image input;
  • text output.

Bro therefore exposes Image-to-Text and image understanding, but does not claim native image generation, transcription, text-to-speech, or audio generation for the installed model.

Execution design

The application separates controls according to the layer they actually affect:

Task Controls
    -> task/system instructions

Response Controls
    -> output contract

Context Controls
    -> message/context construction

Inference Settings
    -> create_chat_completion(...)

Runtime Settings
    -> Llama(...)

This organization keeps the UI aligned with the actual generation functions and prevents task, sampling, context, and runtime settings from being conflated.

Documentation map