Streamline knowledge-intensive business processes with autonomous AI agents. Enterprises face operational bottlenecks in complex, cross-functional processes like onboarding, procure-to-pay, customer claims, and IT operations. These workflows depend on unstructured data, business judgment, and exceptions—areas where traditional automation struggles. AI for Process combines AI reasoning, system integration, and process automation into a unified platform. It enables agentic automation that understands inputs, reasons through decisions, coordinates across systems and teams, and achieves business outcomes autonomously.Documentation Index
Fetch the complete documentation index at: https://koreai.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
Key Components and Capabilities
The platform provides a unified, flexible framework for implementing AI capabilities within an enterprise. It combines pre-built components, customization options, and integration capabilities to address a wide range of business needs while ensuring security and compliance.
| Component | Description |
|---|---|
| Access Points | Entry interfaces — web, mobile, APIs, and direct integrations — for users, systems, and external applications. |
| Application Gateway | Handles request routing, load balancing, and traffic distribution between external access points and internal services. |
| Services | Core AI and business logic layer: Workflow Engine, Expert AI Reasoning, AI model integrations, and Human-in-the-Loop framework. Executes processes, makes decisions, and coordinates multi-agent interactions. |
| Integrations | 250+ pre-built connectors and API adapters for bidirectional data flow with enterprise systems including SharePoint, Salesforce, ServiceNow, databases, and email. |
| Security & Compliance | RBAC, data encryption, audit logging, and compliance frameworks to meet organizational and regulatory requirements. |
| Observability & Monitoring | Metrics, logs, and traces across all components for performance visibility, issue detection, and analytics. |
Key Features
| Feature | Description |
|---|---|
| Workspaces | Logical groupings of workflows and configurations, typically organized by teams or projects. A workspace determines what data and workflows are accessible and which permissions apply to users within that specific environment. |
| Workflows | Structured, executable flows of tasks designed to automate a specific business process. Each workflow run executes independently based on provided inputs and follows a predefined sequence to ensure predictable outcomes. |
| Nodes | Fundamental building blocks of automation arranged and connected on the visual builder canvas. Categorized by function: AI (multimodal processing), Logic (branching and loops), Integration (API calls), and Control (Start/End). |
| Triggers & Schedules | Enable event-driven automation by initiating workflows automatically. Workflows can be triggered by real-time events in third-party apps, predetermined time intervals via the scheduler, external API calls, or manual actions. |
| Context Object | Central runtime state container that captures inputs, intermediate data, and node-level status for a specific execution instance. Dynamically updated as the workflow progresses, allowing nodes to share data through context interpolation using double-curly syntax. |
| Prompt Studio | Dedicated workspace for managing the entire lifecycle of prompts, from creation and testing to optimization. Supports multi-model comparison, version control, and over 65 pre-built templates for common use cases. |
| Model Hub | Centralized interface for full control over the AI model lifecycle. Supports external providers (OpenAI, Google), open-source models (Hugging Face), and fine-tuned models. Allows organizations to customize models using proprietary data and deploy them for inference. |
| Inbox (Human-in-the-Loop) | Centralized queue where human reviewers process tasks that require manual judgment, such as approvals, data input, or content review. The Human Node pauses a workflow until a designated user provides the required input or the task reaches a timeout. |
| Guardrails | Safety measures that deploy scanners to evaluate prompts and AI responses for toxicity, bias, PII anonymization, and prompt injection. Ensure AI interactions remain appropriate and compliant with organizational standards. |
| Observability & Analytics | Suite that includes the Workflow Monitor, Audit Logs, and performance dashboards for visibility into system interactions and efficiency. Tracks metrics such as success rates, average response times, latency, and token consumption. |
| Security & Administration | Enterprise-grade controls that manage users via Role-Based Access Control (RBAC) and secure authentication through Single Sign-On (SSO). Admins can configure Active Directory synchronization to automate user data management. |
| Integrations | Connectivity layer with pre-built connectors for over 120 third-party services, including Salesforce, Slack, and AWS S3. Enables seamless data exchange between the workflow engine and an organization’s existing infrastructure. |
How It Works
┌───────────────────────────────────────────────────────────────────────┐
│ Trigger │
│ (API, Schedule, Event, Manual) │
└──────────────────────────────────┬────────────────────────────────────┘
| ▼ |
┌───────────────────────────────────────────────────────────────────────┐
│ Workflow Engine │
│ │
│ ┌────────┐ ┌────────┐ ┌────────┐ ┌────────┐ ┌────────┐ │
│ │ Start │───►│ AI │───►│ Human │───►│ API │───►│ End │ │
│ │ Node │ │ Node │ │ Node │ │ Node │ │ Node │ │
│ └────────┘ └────┬───┘ └────┬───┘ └────────┘ └────────┘ │
│ ▼ ▼ │
│ ┌───────────┐ ┌───────────┐ │
│ │ Model │ │ Inbox │ │
│ │ Hub │ │ Queue │ │
│ └───────────┘ └───────────┘ │
└──────────────────────────────────┬────────────────────────────────────┘
| ▼ │
┌───────────────────────────────────────────────────────────────────────┐
│ Output │
│ (API Response, File, Notification) │
└───────────────────────────────────────────────────────────────────────┘