File Transformation

Converting Files in raia Academy

Overview

raia Academy enables you to convert virtually any file into an AI‑Ready format—Markdown (.md) or JSON (.json)—before uploading into the vector store.

This ensures your data is clean, structured, and optimized for AI Agent training, regardless of the original file format.

Why AI‑Ready Conversion Matters

Your source data may come from many places and formats:

• Documents: PDF, DOC/DOCX, TXT

• Presentations: PPT/PPTX

• Spreadsheets & Data Files: CSV, XLS/XLSX

• Others: HTML, JSON, Markdown

Each of these formats has its own structure, quirks, and limitations. Without proper conversion, AI models may:

• Miss important text hidden in layouts or tables.

• Misinterpret the structure of the content.

• Produce poor retrieval accuracy due to inconsistent chunking.

• Include “noise” from headers, footers, or formatting artifacts.

By transforming files into clean, structured, and consistent Markdown or JSON, raia Academy ensures:

• Consistent formatting across all knowledge sources.

• Preserved semantic structure (headings, lists, tables).

• High‑quality embeddings for accurate AI Agent responses.

• Reliable chunking so each information block is coherent.

How raia Academy Processes Your Files

1. Upload Your Files

   • Drag & drop files directly into raia Academy.

   • Or bulk import from a shared drive for large-scale processing.

2. AI‑Powered Parsing

   • Automatically detects file type.

   • Extracts unstructured text (narratives, descriptions, notes).

   • Extracts structured data (tables, CSV fields, spreadsheet data).

3. Normalization & Cleaning

   • Removes unnecessary formatting noise.

   • Preserves hierarchy using Markdown (for text-heavy docs) or JSON (for structured datasets).

   • Applies semantic chunking so information is split at logical points.

4. AI‑Ready Output

   • Markdown (.md): Ideal for narrative documents, manuals, and reports.

   • JSON (.json): Ideal for structured data, field/value pairs, and tabular content.

Pushing Files to AI Agents

Pushing to AI Agents

Once files are converted:

1. Select one or more AI Agents in the raia Platform.

2. Push the processed files directly to the Agent’s vector store.

3. Files are immediately available for:

   • Retrieval-Augmented Generation (RAG).

   • Natural language queries.

   • Agent reasoning and training.

Managing Files in raia Academy

• Assign to specific agents so knowledge is targeted.

• Easily remove outdated files from an agent’s vector store to keep responses relevant.

• Re-upload updated versions without disrupting other knowledge sources.

Bulk Upload of Files into raia Academy

Key Benefits

• Format Agnostic – Works with PDFs, Word docs, PowerPoints, spreadsheets, and more.

• Data Consistency – Clean, predictable format improves AI reliability.

• Faster Deployment – Upload → Convert → Push to Agents in minutes.

• Scalable – Bulk process hundreds of files with shared drive import.

• Governance-Friendly – Track, update, and remove files anytime.

Example Workflow

1. Drop a 100-page PDF report into raia Academy.

2. Academy extracts text, tables, and headings → converts to .md.

3. Push the .md to your Market Research AI Agent.

4. The agent can now instantly answer:

   “Summarize the Q4 trends from the latest market report.”

5. Later, update the report → Academy replaces the old version in the vector store.

Best Practices

• Choose Markdown for textual/narrative-heavy content.

• Choose JSON for tabular, database-like, or structured content.

• Keep your files organized by topic before upload for easier agent assignment.

• Use the remove function in Academy to instantly pull outdated knowledge.

Do you want me to make that diagram next?