Can you tell me more about Aivyx's GraphRAG feature?
GraphRAG is a feature employed by Aivyx that aids in memory knowledge graph with BFS traversal, community detection, and agentic RAG. This technology allows Aivyx's AI agents to retrieve, evaluate, and self-correct their context in an encrypted, secure manner, providing an extra layer of security to the user’s data.
Is Aivyx compatible with OpenAI or other AI platforms?
Yes, Aivyx is compatible with leading AI platforms such as OpenAI, Anthropic, or Gemini. It can run entirely on a local machine utilizing Ollama for local inference, or it can connect to these AI platforms whenever more processing power is necessary.
What voice communication features does Aivyx have?
Aivyx supports real-time voice communications through WebSockets. This facilitates interactive and dynamic conversations with the AI agents, which can naturally interpret speech and respond back in real-time. The voice communication feature of Aivyx cannot only interpret spoken words but also reasoning and intent behind the words.
How does Aivyx handle cross-instance collaborations?
To manage cross-instance collaborations, Aivyx utilizes Google's A2A technology for agent-to-agent communication and MCP with OAuth 2.1 for agent-to-tool communication. These sophisticated technologies ensure secure, reliable and efficient communication between agents and tools across different instances.
Can you explain Aivyx's self-improvement feature?
Aivyx includes a self-improvement feature that allows the AI agents to improve over time. This is accomplished through outcome tracking, planner feedback loops, specialist recommendation learning, and skill effectiveness scoring. As the Aivyx's AI agents execute tasks and handle projects, these mechanisms allow them to learn from their actions, feedback, and results to continually enhance their performance and effectiveness.
How can I install Aivyx on my machine?
Installation of Aivyx is made convenient across Linux, macOS, and Windows platforms via a command-line installation procedure. Aivyx also comes with an inbuilt self-update feature that keeps the software current with the most recent updates and features without any manual effort required from the user.
What's in the command-line installation of Aivyx?
Command-line installation of Aivyx is a straightforward process that enables easy setup across various platforms. This installation type is supported on Linux, macOS, and Windows platforms. Once installed, the platform includes a self-update feature, ensuring that users are always on the latest version of the program.
What security measures does Aivyx incorporate for enterprise-level usage?
Aivyx incorporates rigorous enterprise-level security measures, such as Multi-tenancy with per-tenant encryption, RBAC with OIDC SSO, and OWASP alignment. These ensure the necessary security precautions and privacy measures being in place to protect operational data. Aivyx is designed to cater to the specific requirements of large organizations and handle their security needs effectively.
How can Aivyx help in project structures and dependencies analysis?
Aivyx's AI can analyze and review project structures and dependencies, proving to be a robust tool for project management. By breaking down project structures and analyzing dependencies, it aids in identifying potential bottlenecks or inefficiencies, which can then be addressed. The AI utilizes tools and techniques such as GraphRAG for memory knowledge graph with BFS traversal, to evaluate and improve project structure and dependencies.
What features does Aivyx offer for agent-to-agent and agent-to-tool communications?
Aivyx comes incorporated with features for agent-to-agent and agent-to-tool communications. For agent-to-agent communication, it leverages Google's A2A technology. For agent-to-tool interaction, it uses MCP with OAuth 2.1 protocols. These sophisticated communication technologies ensure the smooth operation of tasks and collaborations across different projects and workflows.
What is Aivyx's stance on data ownership and privacy?
Aivyx is firmly committed to data ownership and privacy. The platform was developed with the philosophy of 'own your intelligence' ensuring that user data stays exactly that - user's data. Everything is designed to stay on the user's machine, and all data is encrypted with keys derived from the user's passphrase. Aivyx does not involve any form of telemetry, analytics, or any other means of external data sharing.
What kind of AI agents does Aivyx support?
Aivyx supports specialist agents, each with its own unique identity, tools, and capabilities. Their roles range from Coordinator, Researcher, Analyst, Coder, Reviewer, Writer, Planner, Guardian to Executor. The Coordinator decomposes the user's mission and delegates tasks to the correct specialists, thus ensuring effective project execution and management.
What protocols does Aivyx use for collaboration?
For collaboration protocols, Aivyx uses Google's A2A for agent-to-agent communication and MCP with OAuth 2.1 for agent-to-tool communication. Forestation is used for cross-instance collaboration. The use of industry-standard interoperable protocols ensures the smooth flow and exchange of information, leading to efficient collaboration.
Can Aivyx run locally or does it require cloud connections?
Yes, Aivyx is designed to run locally on a user's machine. It doesn't require any cloud connection as all operations can be performed offline, hence emphasizing its privacy-first approach. However, for users seeking more power, Aivyx can connect to AI platforms such as OpenAI, Anthropic, or Gemini.
What is Aivyx?
Aivyx is an open-source AI platform that prioritizes privacy and data ownership for its users. The agent performs its functions entirely on the user's machine, thus ensuring encryption of data at rest. It offers significant features like 22 built-in skills, local functionality for offline use, capability to connect to external tools via Model Context Protocol, and a tamper-proof audit log. With 9 specific AI agents known as 'The Nonagon', Aivyx caters to diverse workflow styles. It provides a pleasing terminal UI and a native desktop app compatible with Linux and Windows for easy operation.
How does Aivyx protect my data?
Aivyx implements a vigilant approach to data protection. It uses encryption measures like ChaCha20Poly1305 and HKDF-SHA256 to encrypt the user's data while at rest. The agent runs entirely on a user's local device, and all stored data remains on the device, cutting off any possibility of telemetry, analytics, or phone-home. It also doesn't store the passphrase, only keeping derived keys in memory, which are zeroed on process exit.
What kind of workflows can Aivyx handle?
Aivyx can handle a myriad of workflows due to its robust feature set. Parallel step execution with dependency graphs allows multiple activities to be conducted seamlessly, and reflection loops with the 'LLM-as-judge' function enable human-in-the-loop approval checkpoints. The Nonagon, a team of 9 specialized AI agents, further expands workflow capabilities, catering to different tasks such as research, code review, writing, analysis, etc. For enterprise users, Aivyx can manage multi-tenant isolation, compliance, cost control, and more.
What operating systems is Aivyx compatible with?
Aivyx is compatible with multiple operating systems including Linux, macOS, and Windows. The selection of install packages makes the installation process easy, with one-line installation commands provided for Linux, macOS, and Windows alongside Flatpak, Homebrew, Snap, and AUR packages. It also includes a built-in `aivyx self-update` command.
Could you elaborate on the 22 built-in skills Aivyx has?
The specific details regarding the 22 built-in skills of Aivyx are not explicitly listed. However, potential skills could include functionalities related to the tasks that the specialized AI agents ('The Nonagon') handle, such as code review, research, writing, and analysis. It is reasonable to infer that the skills allow Aivyx to perform a wide range of tasks beneficial to professional workflows.
What is the Model Context Protocol Aivyx uses to connect to external tools?
The Model Context Protocol in Aivyx enables connectivity to external tools. While the specific workings of this protocol have not been detailed, its purpose is clear. It provides a way for Aivyx to interface and communicate with external applications and services, thus enhancing the AI's functionality and ability to incorporate a wider range of services and tools.
What is the tamper-proof audit log feature in Aivyx?
Aivyx's tamper-proof audit log feature is implemented using an HMAC chain which records every action taken. The purpose of this feature is to provide a secure, verifiable record of all activities performed by the AI. Owing to the HMAC chain employed, the audit log is tamper-proof, ensuring the integrity and reliability of the logged data.
Could you explain more about the specialized AI agents ('The Nonagon') in Aivyx?
'The Nonagon' refers to 9 specialized AI agents, each with unique identities, abilities, and tools. These AI agents work together, handling different functions to provide a comprehensive AI experience. Each agent has tasks it specializes in, contributing to a holistic, effective, and productive workflow. However, more specific details about The Nonagon aren't available.
What data encryption measures does Aivyx use?
Aivyx uses robust encryption measures for all data. The agent employs ChaCha20Poly1305 authenticated encryption to secure data at rest. Additionally, it uses HKDF-SHA256 to derive keys for this encryption process. Data remains encrypted and local to the user's machine, enhancing the security and privacy of the AI user experience.
Does Aivyx support real-time streaming?
Yes, Aivyx supports real-time streaming. It incorporates real-time voice functionality achievable through WebSocket technology. This setup gives rise to functionalities like speech-to-text, agent reasoning with tools, and text-to-speech streaming.
What data types can Aivyx extract from?
Aivyx has been built to extract data from a range of file types. It can understand and extract data from PDF files, XLSX files, and CSV files. This feature makes Aivyx versatile in handling and processing data from common file formats.
What is the HKDF-SHA256 Encryption key in Aivyx?
HKDF-SHA256 in Aivyx is an encryption mechanism used to derive encryption keys. It is a part of Aivyx's security protocol where all data is encrypted at rest. The HKDF-SHA256 algorithm is crucial in deriving keys to encrypt this data, and it helps in guaranteeing the security and privacy of user data.
How does Aivyx's self-improvement mechanism work?
Aivyx's self-improvement mechanism incorporates outcome tracking, feedback loops, specialist recommendation learning, and skill effectiveness scoring. This system allows the AI agents to improve their performance over time by learning from their actions, outcomes, and feedback, rendering the AI experience progressively better and more effective for users.
What languages was Aivyx written in?
Aivyx has been written in Rust programming language. Rust is known for its efficiency and control of low-level system details making it fitting for developing performance-centric applications like Aivyx.
What is the outcome tracking feature in Aivyx?
Outcome tracking in Aivyx is a component of its self-improvement mechanism. This feature allows the AI to monitor the results of its actions, learn from them, and adjust its behavior accordingly. The intent is to continuously fine-tune the AI’s behavior to better serve user needs and optimize its performance.
Can Aivyx operate offline?
Yes, Aivyx can operate offline. It has a local-first design that allows it to run entirely on a user's machine with the help of Ollama for fully local inference. This means Aivyx can continue to function even without an internet connection. When necessary, it can connect to OpenAI, Anthropic, or Gemini for more power.
How does Aivyx handle encryption and decryption of data?
Aivyx uses robust encryption measures, such as ChaCha20Poly1305 and HKDF-SHA256, to encrypt data at rest. It never stores the user's passphrase, only using it to derive encryption keys that are stored in memory. When the process exits, all key material is removed. Thus, all data remains encrypted and unreadable without the user's passphrase, ensuring data security.
What is the 'voice agent' feature in Aivyx?
The 'Voice Agent' feature in Aivyx allows users to interact with the AI using speech. This is implemented using WebSocket technology for real-time communication, including speech-to-text conversion, agent reasoning with tools, and streaming of text-to-speech responses. Aivyx allows for natural conversation and even supports barge-in interruptions.
What security features does Aivyx offer to enterprise users?
Aivyx offers comprehensive security features for enterprise users. It provides multi-tenancy with per-tenant encryption, meaning each tenant's data is separated and encrypted differently. It also offers RBAC with OIDC SSO for access controls, cost governance with budgets for finance management, and prompt injection defense for additional security. Aivyx's security measures align with OWASP guidelines, providing enterprise-level security controls.
Can Aivyx connect to OpenAI and other platforms if needed?
Yes, Aivyx can connect to other platforms like OpenAI, Anthropic, or Gemini if required. While Aivyx can run completely locally using Ollama, when users need more power for complex tasks, it has the capability to connect to these platforms and leverage their functionalities.