Talk to My Docs¶

GitHubでこのアプリケーションテンプレートにアクセス

Use the Talk to My Docs application template to build, develop, and deploy an AI-powered application that allows you to dynamically talk to your documents across different providers, such as Google Drive, Box, and your local computer using multi-agent orchestration, a modern web frontend, and robust infrastructure-as-code.

このアプリケーションは、以下のような場合に役立ちます。

• Generate reports and dashboards for stakeholders and operationalize work from documents generated by data analysts, data scientists, etc.

• 各部門の業務、不定期な分析依頼、定期報告書の作成と自動化について、パフォーマンスを追跡、報告、分析する。

• ​​Empower faster decision-making by rapidly extracting insights from your documents.

• デプロイのセキュリティとガバナンスを制御する。

• 企業のニーズに合わせて業務の規模を調整する。

主な機能¶

• Pre-built AI assistants: For instant productivity, choose from a gallery of specialized assistants, including:

  • Contract Review Bot
  • HR Policy Assistant
  • Feedback Bot

• Document catalog and search: Quickly find and reuse previously uploaded documents across all your knowledge bases and conversations. With Talk to My Docs, enterprises get the productivity of consumer AI tools with the security and governance they require.

• Supports multiple document types: Upload PDFs, Word docs, spreadsheets, and text files to the application.
• Knowledge bases for team collaboration: Create shared document collections that teams can search, analyze, and build upon together.
• Secure enterprise data source integration: Connect to Google Drive and Box with secure authentication and RBAC controls.
• Natural language-powered, context-aware Q&A with enterprise security: Use everyday language and easy prompts to ask questions about your documents, leveraging your proprietary datasets for precise answers in a conversational chat experience with minimal security risks. Continue conversations across sessions, add new documents to existing chats, and maintain full context for deeper analysis.
• Enterprise governance: Full RBAC controls, audit trails, and usage monitoring to ensure secure, compliant AI deployments.

ユースケース¶

Review some use cases that are suited for using Talk to My Docs:

• Invoice and receipt processing: Businesses that handle a large number of invoices, receipts, and purchase orders can quickly extract details like invoice numbers, line items, and vendor information, making it easier to review specific information, assess risk, and check compliance.

• Financial analysis: Automate data extraction for analysis, fraud detection, and risk assessment by parsing various documents, including bank statements, loan applications, credit reports, and financial reports.

• Customer sentiment analysis: Understand customer sentiment by uploading customer reviews, social media mentions, feedback forms, etc., to identify trends, improve products and services, and enhance customer satisfaction.

アーキテクチャ¶

各テンプレートは、元データの入力からデプロイされたアプリケーションまで、エンドツーエンドのAIアーキテクチャを提供する一方で、特定のビジネス要件に合わせて高度なカスタマイズが可能です。

アプリテンプレートには、3つの補完ロジック群があります。 このテンプレートでは、完全にカスタムのAIロジックとフロントエンドを選択するか、DataRobotの既存の機能を利用することができます。

• AIロジック：AIリクエストへの対応、予測の生成、予測モデルの管理に必要です。

• アプリロジック：ホストされたフロントエンドを経由する場合でも、外部の消費レイヤーに統合する場合でも、ユーザーの利用に必要です。

• 運用ロジック：すべてのDataRobotアセットを有効にするために必要です。