Skip to content

[Feature]: Enable AI Vision and PDF Parsing for Knowledge Base Uploads #249

Description

@daniilperkin

User Story

As a user, I want to upload images and PDFs to the knowledge base, so that the AI can seamlessly caption, parse, and summarize their contents.

Context & Motivation

Previously, the platform was optimized for plain text and source code. While the UI permitted binary uploads, they silently failed to summarize because the data pipeline dropped the binary content before it reached the AI parsing engine.

This issue ensures end-to-end support:

Frontend: Accepting the files in the UI.
Backend: Encoding the binary payload for safe transit.
AI Service: Decoding the payload, parsing PDFs, and captioning images.

Acceptance Criteria

Users can upload images (png, jpg, webp) and PDFs via the knowledge base UI.
Uploaded binary files are properly Base64 encoded and dispatched by the backend.
The AI service processes Base64 payloads, extracting text from PDFs and generating rich captions for images.
Users see a meaningful AI-generated summary for uploaded binary files.

Technical Notes

Frontend: [Feature]: Support image and PDF selection in Knowledge Base upload UI (Already completed)
Backend: [Feature]: Pass binary file contents (Images, PDFs) to AI service during batch upload ingestion
AI Service: [Feature]: Add vision captioning and Base64 PDF decoding to batch ingestion endpoint

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions