Agent Cloud

Agent Cloud is sunset and no longer active.

What was Agent Cloud?

Agent Cloud was an open-source platform designed to empower companies in building and deploying private Large Language Model (LLM) chat applications. It enabled teams to securely interact with their data, offering a robust solution for various AI-driven applications.

Key Features and Functionality:

• Data Synchronization for Vector Databases: Agent Cloud facilitated the creation of data-connected agents with access to up-to-date vector data for Retrieval Augmented Generation (RAG).
• Versatile Data Ingestion: The platform supported the ingestion of data from over 260 sources, including files (PDF, DOCX, TXT, CSV) and databases (PostgreSQL, Snowflake, BigQuery), with options to select specific tables and columns.
• Data Preparation: Agent Cloud allowed users to split, chunk, and embed data with instructions, leveraging models like OpenAI's text-embedding-3-small or open-source alternatives like BGE/base.
• Vector Data Storage: Embedded data was stored within a vector database, supporting major options like Qdrant and Pinecone for self-hosted deployments.
• Scheduled Data Updates: Users could set the frequency for data synchronization from the source, ensuring access to the freshest data via manual, scheduled, or cron-based updates.
• LLM Agnostic Design: Agent Cloud was designed to be LLM agnostic, allowing users to connect their own open-source models or leverage OpenAI. Self-managed users could connect to locally hosted models for full privacy.

How Agent Cloud Worked:

1. Data Source Selection: Users selected their data sources from a collection of connectors, including systems like Confluence or direct file uploads.
2. Data Preparation: Instructions were provided on how to split and chunk data, with options to use OpenAI's text-embedding-3-small or open-source models for embedding.
3. Vector Storage: The platform stored the embedded data in a vector database.
4. Data Synchronization: Users set a frequency to sync data from the source.
5. Chat Interface: Agents were created with the user's choice of LLM, enabling chat sessions with the synced data.

Use Cases:

• Building RAG Chat Applications: Agent Cloud enabled the creation of chat applications for querying synced data directly through a chat interface or APIs.
• Private Data Interaction: It allowed teams to securely access AI apps within their organization, ensuring data privacy by deploying the open-source app on their infrastructure and running LLMs on-premise.

Who Was Agent Cloud For?

Agent Cloud was aimed at developers and engineers looking to build AI applications for themselves and their teams, providing a platform for creating single-agent chat apps, multi-agent chat apps, and knowledge retrieval apps.

Agent Cloud has been sunset, but the site remains online for archival purposes.