# Abacus AI MCP for AI Agents AI Agent Connect > Abacus AI (Enterprise AI Cloud) MCP lets you manage your entire machine learning lifecycle through your AI agent. Instead of jumping between tabs, you can create projects, manage datasets, kick off training jobs, and deploy models to real-time endpoints using natural language. It turns your AI client into a command center for MLOps. ## Overview - **Category:** cloud-infrastructure - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_Lr50dCQPDxtMBNSIYHpuANfOgvy9zpJYnqAO4ozi/ai-agent-connect - **Tags:** machine-learning, mlops, model-training, predictive-analytics, enterprise-ai ## Description Abacus AI (Enterprise AI Cloud) gives you a way to manage your entire machine learning lifecycle through your AI agent. Instead of jumping between different web consoles, terminal windows, and documentation pages, you can handle your MLOps tasks in one place. This Connector bridges the gap between your workspace and the Abacus AI Enterprise Cloud, letting you orchestrate complex workflows using just natural language. You can create new projects, organize your datasets, and kick off training jobs without ever leaving your current environment. When you connect this to your workflow via the Vinkius catalog, you stop worrying about remembering specific IDs or navigating deep menus. You can check the health of your models, deploy them to live endpoints, and pull predictions in real-time. It turns your AI client into a command center for your ML infrastructure. You can verify dataset metadata to ensure your data is ready for training or initiate a training job with specific configurations by simply describing what you want to do. Whether you're a data scientist checking a training status or an engineer pushing a new model to production, this tool removes the friction of manual management. It handles the heavy lifting of project orchestration and deployment so you can focus on the actual science. You get a streamlined way to move from raw data to live inference, keeping everything organized and accessible within your primary workspace. ## Tools ### create_deployment Push a trained model to a real-time endpoint for inference. This moves your model from a static state to a live production environment. ### describe_dataset View the metadata and structure of an existing dataset. Use this to understand the underlying data before starting a project. ### describe_model Get the current status and specific details of a trained model. Use this to see if your training job finished successfully. ### get_prediction Fetch a live prediction from a deployed model using your input data. This is great for testing how your model handles specific user inputs. ### list_projects See every project currently in your organization. Use this to get a high-level view of all active ML work. ### train_model Start a new training job with your desired configurations. You can specify the parameters for your model directly through the agent. ### create_project Create a new machine learning project in your Abacus AI account. This helps you organize your work into specific use cases. ### create_dataset Build a new dataset for your machine learning projects. This helps you organize your data before you begin training. ## Prompt Examples **Prompt:** ``` List all my machine learning projects in Abacus AI. ``` **Response:** ``` I've retrieved your projects. You have 3 active projects: **Customer Churn** (ID: proj_123), **Demand Forecast** (ID: proj_456), and **Lead Scoring** (ID: proj_789). ``` **Prompt:** ``` Start training a model named 'SalesModel_v1' in project proj_123. ``` **Response:** ``` The training job for **SalesModel_v1** has been initiated in project proj_123. You can check its progress using the model ID provided: **model_abc123**. ``` **Prompt:** ``` Get a prediction from deployment dep_999 using this data: {"user_id": "123", "last_purchase": 50}. ``` **Response:** ``` The model predicts a **churn_probability** of **0.15** for this user based on the provided data. ``` ## Capabilities ### Create new ML projects Start new projects for specific use cases like churn prediction or retail recommendations. ### Build and inspect datasets Create datasets and check their metadata to understand data structures before training. ### Start model training Initiate training jobs with custom configurations using natural language commands. ### Deploy to live endpoints Push trained models to real-time endpoints for immediate inference. ### Get real-time predictions Retrieve instant predictions from deployed models using your own input data. ### Monitor model health Query the metrics and status of your models and datasets to keep your infrastructure running. ## Use Cases ### Checking training progress An ML engineer needs to see if a model is finished. They ask the agent for the status of a specific model ID and get the result immediately. ### Verifying dataset structures A data scientist needs to see what's in a dataset. They use the agent to describe the metadata of a specific dataset to ensure it is ready for training. ### Monitoring active projects A product manager wants to see all active projects. They ask the agent to list all projects in the organization to get a high-level status update. ### Testing live predictions A developer needs to test a new deployment. They provide JSON data to the agent to get a prediction from a specific deployment to verify the output. ## Benefits - Stop switching tabs by using list_projects and describe_model to see your work status instantly in your workspace. - Speed up deployment cycles by using create_deployment to push models to production without manual cloud configuration. - Get faster data insights by using describe_dataset to check metadata directly in your chat without opening a browser. - Reduce human error in training by using train_model to initiate jobs with specific configurations via natural language. - Simplify inference testing by using get_prediction to see how your model handles real data in seconds. - Keep your infrastructure running by using describe_model to query the health and metrics of your models on demand. ## How It Works The bottom line is you get a natural language interface for your entire ML pipeline. 1. Connect your Abacus AI API key to the Connector via Vinkius. 2. Tell your AI agent to perform a specific ML task like starting a training job. 3. The agent executes the command and gives you the status or results immediately. ## Frequently Asked Questions **Can I use the Abacus AI MCP to manage my ML projects?** Yes, you can list, create, and describe ML projects using natural language. This lets you keep your project organization tidy without leaving your workspace. **Does the Abacus AI MCP support model training?** It lets you start training jobs and check their status without leaving your chat. You can initiate jobs with specific configurations just by asking your agent. **How do I get predictions using the Abacus AI MCP?** You can ask your agent to pull a live prediction from a specific deployment using your input data. This is a fast way to test your model's performance on real-world inputs. **Can I use this to see my dataset metadata?** Yes, the Connector allows your agent to inspect dataset structures so you know exactly what data you're training on before you start a job. **Is the Abacus AI MCP good for MLOps?** It's built specifically for MLOps, handling the transition from training to deployment and monitoring. It removes the manual work of moving models through the pipeline. **Can I deploy models to endpoints with this Connector?** Yes, you can use it to push trained models to real-time endpoints and manage those deployments through your AI client. **How can I check if my model training is finished?** You can use the `describe_model` tool by providing the unique Model ID. It will return the current status, metrics, and other details of the training job. **Can I get a prediction from a deployed model directly through the agent?** Yes! Use the `get_prediction` tool. You will need the deployment ID, the deployment token, and the input data in JSON format to receive a real-time prediction. **Is it possible to create a new project for a specific ML use case?** Absolutely. Use the `create_project` tool and specify the name and the `useCase` (e.g., 'RETAIL_RECOMMENDATIONS') to initialize a project tailored for that specific application.