Agent Cloud

Data sync for Vector DBs

Freemium

Description

Agent Cloud provides a comprehensive solution for creating data-connected AI agents that leverage Retrieval-Augmented Generation (RAG) with fresh, up-to-date vector data. It facilitates the entire RAG pipeline, allowing users to connect to over 260 data sources, including systems like Confluence, databases (Postgres, Snowflake, BigQuery), and direct file uploads (PDF, DOCX, TXT, CSV, XLSX). The platform includes a built-in data pipeline for splitting, chunking (using methods like character splitting or semantic chunking), and embedding data into vector stores, ensuring data is prepared optimally for AI consumption.

Designed with flexibility and security in mind, Agent Cloud is LLM agnostic, supporting connections to both open-source models (like locally hosted Ollama or LM Studio via OpenAI compatible endpoints) and cloud-based models such as OpenAI and Azure OpenAI. It integrates with major vector databases like Qdrant and Pinecone, particularly beneficial for self-hosted deployments. Available as an open-source community edition (AGPL 3.0) for self-hosting or as a managed cloud service (details not fully provided), Agent Cloud emphasizes private and secure AI application deployment within organizations, enabling teams to safely access and interact with company data through AI agents.

Key Features

Data Synchronization: Keeps vector data updated via manual, scheduled, or cron-based syncs for RAG.
Extensive Data Source Integration: Connects to 260+ data sources (e.g., Confluence, Postgres, Snowflake, BigQuery) and supports file uploads (PDF, DOCX, TXT, CSV, XLSX).
Built-in Data Pipeline: Manages data splitting (character/semantic), chunking, and embedding.
Flexible Embedding Options: Supports OpenAI embedding models and local open-source models via fastembed.
LLM Agnostic: Compatible with open-source LLMs (via OpenAI compatible endpoint) and cloud models (OpenAI, Azure OpenAI).
Vector Database Support: Integrates with major vector DBs like Qdrant and Pinecone.
Secure Access Management: Controls data access for LLM apps and organizational teams.
Open Source Option: Offers a community edition (AGPL 3.0) for self-hosting.
API Access: Enables building custom chat applications and integrations.
Customizable Ingestion: Control over synced fields and designation for embedding vs. metadata.

Use Cases

Building Retrieval-Augmented Generation (RAG) chat applications.
Creating AI agents with access to up-to-date company knowledge.
Developing secure internal AI tools for various departments (Sales, IT, HR, Support, Legal).
Deploying private and secure AI applications on proprietary infrastructure.
Building knowledge retrieval systems connected to diverse data sources.
Enabling AI agents to interact with synced vector data.

Frequently Asked Questions

What is your software license?

AGPL 3.0 - it is a copy left license which can be found on our github page.

What hardware requirements do I need to run Agent Cloud locally?

If running via Docker, we strongly recommend a machine with at least 16 GB of RAM. A base Macbook Air M1/M2 with 8GB RAM will not suffice. Additional RAM is needed if running local LLM models like Olama or LM Studio.

Can I use a local Large Language Model?

Yes. We support any local LLM which has an Open AI compatible endpoint (i.e., it responds the same way Open AI does). This includes LM Studio or Ollama.

What splitting/chunking methods do you support?

For files, we support Basic (character splitting) and Advanced (semantic chunking). For other data sources, chunking currently happens per message (e.g., row by row for BigQuery), with plans for more granular control.

How can my data retrieval be truly private?

For true privacy, deploy our open source app to your infrastructure, run an LLM on-premise or in your own cloud, and consider an enterprise self-managed license.