AutoArk: A Multi-Intelligence AI Platform that Collaborates on Complex Tasks

AutoArk is a company focusing on artificial intelligence technology, the core of which is the self-developed end-to-end multimodal model EVA-1, which outperforms GPT-4o in a number of international benchmarks. Based on the EVA-1 model, AutoArk has further developed a multimodal framework called "ArkAgentOS Based on the EVA-1 model, Boundless Ark has further developed a multi-intelligence framework called "ArkAgentOS". The design goal of this framework is to allow multiple AI intelligences to work together as a team to solve complex tasks that are difficult for a single AI to handle. It lowers the barriers to using advanced AI technology by mimicking the mindset of industry experts and enabling users to command AI systems in natural language to automatically generate and adjust workflows. The company's technology focuses on solving two problems: first, how to enable AI to handle highly specialized and complex tasks with complex processes, such as in the fields of biomedical research and development or writing professional consulting reports; and second, how to enable AI to provide personalized services while protecting users' private data. Currently, Boundless Ark has launched two main products: one is the "AgentStudio" platform for individuals and enterprises, which is used to create and manage intelligent bodies that can perform work automatically; the other is the AI desktop robot called "Arki", which focuses on life companionship and daily life. The other is an AI desktop robot called "Arki", which focuses on life companionship and daily assistance.

Function List

• EVA-1 multimodal macromodel. As a technological base, the model is able to integrate text, images, audio and other information to achieve a smarter and more anthropomorphic interactive experience.
• Multi-Intelligence Framework (ArkAgentOS). Allows the creation of multiple intelligences that can divide up the work and work together to accomplish a complex total task.
• Workflow automation. The system automatically generates and orchestrates workflows by simply having users talk in natural language, eliminating the need to manually design the process.
• Unsupervised Training. AI systems can learn and evolve themselves as they perform tasks and collect data, automatically adapting to new and unknown workflows.
• End-to-end multimodal capabilities. Based on the EVA-1 model, it supports multiple forms of inputs and outputs such as voice, text, images, etc. The delay of voice dialog is as low as 400 milliseconds, and it supports interjections and interruptions at any time during the dialog.
• AI Emotional Systems. The EVA-1 model has the ability to recognize and express emotions, allowing for more humane interactions.
• AgentStudio Platform. Provides a visualization platform that integrates memory, tool invocation, and workflow generation for building and managing multi-intelligencer applications powered by EVA-1.
• Arki AI Desktop Robot. A solid robot based on the EVA-1 multimodal macromodel capable of providing emotional companionship and life assistance functions.
• Privacy-preserving technologies. The first "Privacy Patch" technology, utilizing a distributed "big and small brain" architecture, can provide services without accessing users' explicit data.
• Hardware drivers. AI models can directly control and drive 3D avatars or external hardware devices.

Using Help

AutoArk provides a software platform called "AgentStudio" and a hardware robot called "Arki", both of which are based on its self-developed EVA-1 multimodal model.

AgentStudio: A One-Stop Multi-Intelligent Body Task Solution Platform

AgentStudio is a platform that allows users to create, manage and collaborate with multiple AI intelligences to automate complex tasks. It encapsulates the power of the EVA-1 Big Model into an application that can be understood and operated by the average user.

How to get started with AgentStudio:

The use of the platform is expected to be based on the mainstream SaaS (Software as a Service) model, where users can register for an account and log in through the official website.

1. Login and access to the workspace.
   • interviewsautoarkai.comTo do so, find the AgentStudio portal and log in to your account.
   • When you enter the main interface, you will see a "workspace" or "project dashboard". This will show you the projects and workflows you have created.
2. Create the first workflow (core functionality).
   • Create with Intent-Driven. The core feature of AgentStudio is natural language driven. Instead of dragging and dropping complex flowcharts, you can enter your end goal directly into a dialog box.
   • For example, you can type, "Help me write an industry analysis report on New Energy Vehicle Market 2024, which needs to include four sections: market size, key players, technology trends and future forecasts."
   • AI auto-sequencing. Upon submitting this command, the platform's built-in, EVA-1-powered AI Copilot (Intelligent Copilot) will get to work. It will automatically break down this complex task into a series of sub-tasks and assign the right intelligence to each of them.
     • For example, it may generate a workflow that includes "data collection intelligence," "data analysis intelligence," "report writing intelligence," and "chart generation intelligence. report-writing intelligence" and "chart-generating intelligence".
   • Human-Machine Synergy and Fine-Tuning. After AI generates the workflow, it is presented to you in a visualization. You can check each step and make changes. For example, you can tell the AI: "In the data collection section, please focus on China market data" or "The tone of the report writing intelligence needs to be more formal". This human-machine collaboration ensures that the end result meets your precise requirements.
3. Intelligent body specific operations:
   • Configuration Intelligence. In the workflow, you can click on each intelligence to configure it in detail. For example, for the Data Collection Intelligence, you can specify its sources of information, such as limiting it to specific news sites or academic databases.
   • Calling Tools. Intelligentsia can use a variety of "tools" to accomplish tasks such as联网搜索,文件读写,API调用etc. You can authorize or restrict which tools they use in the configuration.
   • Long- and short-term memory. The platform provides memory capabilities for intelligences. They remember your previous conversations, preferences, and project history, making collaboration more consistent.
4. Implementation and monitoring.
   • When the configuration is complete, click the "Start Execution" button.
   • You can see in real time on the dashboard the progress of the execution of the entire workflow, which intelligences are working, which have finished, and the process of transferring information between them.
   • If something goes wrong or the result is unsatisfactory, you can pause the task, adjust the instructions or configuration for that intelligence, and then resume execution.
5. Result Output.
   • Once all the intelligences have completed their work, the system integrates the final result. Using the previous industry report as an example, you'll end up with a complete document containing text, data, and charts, all generated collaboratively by a team of AI intelligences.

Arki: AI Desktop Robot

Arki is a physical hardware product whose interaction experience directly benefits from the end-to-end multimodal capabilities of the EVA-1 macromodel.

1. Unboxing & Setup.
   • Connect the Arki robot to a power source and turn it on.
   • Follow the instructions and use the mobile app to scan the QR code on the device to connect Arki to your home Wi-Fi network.
   • Log in to your Boundless Ark account in the app and complete the initial setup, such as choosing the robot's voice, language preferences, etc.
2. Daily Interaction.
   • Wake Up Call. Arki uses voice wake-up, so all you have to do is say the wake-up word (e.g., "Hello, Arki") and its screen or light will light up to indicate that it's listening.
   • Natural Conversations. You can talk to it directly. Thanks to the low-latency technology of the EVA-1 model (400 ms response), the conversation will be very smooth and close to human-to-human communication. You can interrupt it at any time and it will respond instantly.
   • Emotional Companionship. Arki has EVA-1's built-in Emotion System, which provides emotional support by displaying different expressions (e.g., happy, curious) on the screen based on your tone of voice and wording, and answering you with a tone of voice that carries the corresponding emotion.
   • Assisted Living. You can ask it for information, such as "What's the weather like today?", "Help me set an alarm for 8 am tomorrow", or do unit conversions and look up recipes. or "Set an alarm for 8 am tomorrow", or do unit conversions, look up recipes, and more.
3. Multimodal Functions.
   • In addition to voice, Arki may also be equipped with a camera. You can show it objects and ask it questions, such as picking up an apple and asking it, "How many calories are in this?" .
   • It can also drive its own physical parts to make some simple movements in response to your commands, enhancing the fun of interaction.

application scenario

1. Biomedical R&D
   Researchers can instruct the AI to automatically collect all the latest papers on a specific target, summarize and organize them by the data analysis intelligences, and then write the literature review by the report generation intelligences, and the whole process is automated to complete, dramatically shortening the data preparation time in the pre-research and development stage.
2. Market Analysis and Consulting
   Analysts in corporate marketing departments or consulting firms can use AgentStudio to input commands such as "Generate a consumer insight report on the coffee machine market" and let the AI automatically complete data capture, competitor analysis, sentiment analysis of user reviews, and report writing to quickly produce high-quality business decision-making basis.
3. Personal life and emotional accompaniment
   The Arki desktop robot can be used by people who live alone or need companionship. Users can engage in daily chit-chat with Arki and share their moods, and Arki is able to respond with emotion through the built-in emotion system of its EVA-1 model, making it an intelligent "electronic pet" or life companion to alleviate the feeling of loneliness.
4. Enterprise automation office
   Enterprises can use AgentStudio to automate daily office tasks, such as automatically reading newly received customer emails, determining the intent of the email and distributing it to the appropriate person in charge by AI, as well as automatically creating task cards and setting deadlines in the project management software.

QA

1. How is Boundless Ark's AI different from other big models?
   The main difference is that it is centered on the self-developed EVA-1 end-to-end multimodal macromodel and the "multi-intelligence" collaborative framework (ArkAgentOS) based on this model. An ordinary big model is like a knowledgeable "generalist", but it can only process one command at a time. Boundless Ark's AI is like a team of multiple "specialists" that can work together to handle longer and more complex task chains, especially for specialized workflows that require multiple steps to complete.
2. What are the specific advantages of the EVA-1 model?
   EVA-1 is a large multimodal model benchmarked against OpenAI GPT-4o. Its core advantages include: 1) end-to-end multimodal processing capability to understand and generate text, image, audio and other information simultaneously; 2) extremely low interaction latency and fast response to voice dialog; 3) built-in emotion perception and expression capability for more anthropomorphic interaction; and 4) powerful professional reasoning capability and ultra-long text comprehension.
3. Is programming knowledge required to use AgentStudio?
   No need. The core design concept of the platform is "natural language-driven". Users don't need to write code, they just need to describe their goals and requirements in everyday language as if they were having a conversation with a human being, and the AI system will automatically orchestrate and execute the workflow. This makes it accessible to non-technical professionals.
4. How does the platform protect my data privacy?
   Boundless Ark pioneered the "Privacy Patch" technology. It adopts a distributed architecture of "big brain and small brain", where sensitive data is left in the local "small brain" to be processed and generate an unresolvable "intermediate state" (e.g., a string of digits) before being uploaded to the cloud "brain" for computation. This intermediate state is then uploaded to the cloud "brain" for computation, and the original text and plaintext data do not leave the user's device, thus protecting privacy and security while providing powerful AI services.