Skip to main content

Platform Availability and Access

The Agent Platform is available on our US Cloud at https://agent-platform.kore.ai. To request access for your customers, partners, or prospects, contact Support.

Outbound IP Addresses

The Platform sends requests from a fixed set of IP addresses. If your systems use IP allowlisting, add these IPs to your allowlist so the Platform’s traffic is permitted. Outbound IP Addresses (SaaS instance: https://agent-platform.kore.ai):
  • 34.201.194.187
  • 44.216.129.184
  • 54.225.127.87

Agentic App Setup Guide

Set up your Agentic App in four phases:
┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │ 1. Foundation │ ──► │ 2. Configuration │ ──► │ 3. QA & Test │ ──► │ 4. Deployment │ │ │ │ │ │ │ │ │ │ Account & │ │ Agents, tools, │ │ Validate & │ │ Go live & │ │ core setup │ │ & knowledge │ │ test safely │ │ integrate │ └─────────────────┘ └──────────────────┘ └─────────────────┘ └─────────────────┘

Phase 1: Foundation

Get your workspace and core components ready.
StepDescription
1. Sign UpCreate an account to get a dedicated workspace. If you’ve used XO Platform, your existing workspaces appear automatically. To use a shared workspace, you’ll need permission from its owner. Learn more →
2. Configure an LLMChoose and connect a large language model (LLM) — the intelligence that powers your agents. Supported providers: OpenAI, Azure OpenAI, Anthropic, Gemini. Learn more →
3. Create an Agentic AppStart from scratch, import a pre-built app, or install one from the marketplace. Learn more →

Phase 2: Configuration

Build out your app’s agents, knowledge, and tools.
StepDescription
1. Set Up AgentsDefine each agent’s purpose and assign the tools and resources it needs to complete tasks. Learn more →
2. Add KnowledgeConnect knowledge sources so agents can retrieve accurate, up-to-date answers. Sources can include FAQs, documents, databases, and external systems. The Platform uses RAG to search them. Learn more →
3. Configure ToolsTools let agents take action — fetching data, running logic, and triggering workflows. See the tool type matrix below to pick the right tool for your use case. Learn more → Workflow Tools, Code Tools, MCP Tools
4. Configure EventsEvents mark key moments in a conversation: welcome, agent handoff, end of session. Customize how your app handles each to keep interactions smooth and consistent. Learn more →
Which tool type should you use?
┌─────────────────┬─────────────────────────────────────────────────────────┐ │ Tool Type │ Use When… │ ├─────────────────┼─────────────────────────────────────────────────────────┤ │ Workflow Tool │ You want a no-code visual builder for step-by-step │ │ │ processes. │ ├─────────────────┼─────────────────────────────────────────────────────────┤ │ Code Tool │ You need custom logic using JavaScript or Python. │ ├─────────────────┼─────────────────────────────────────────────────────────┤ │ MCP Tool │ You want a universal connector to external tools │ │ │ and services via the Model Context Protocol. │ └─────────────────┴─────────────────────────────────────────────────────────┘

Phase 3: Quality Assurance

Validate safety, accuracy, and performance before going live.
StepDescription
1. Deploy GuardrailsAdd input and output scanners to control what agents receive and send. This prevents unsafe, off-topic, or unexpected responses. Learn more →
2. Test the AppUse the built-in simulator to submit queries and review responses. Catch errors and edge cases before your users do. Learn more →
3. Evaluate PerformanceRun structured evaluations in Evaluation Studio to measure response quality, spot coordination issues, and optimize agent behavior at scale. Learn more →

Phase 4: Deployment

Go live and connect your app to your products.
StepDescription
1. Deploy the AppPromote your tested app version to staging or production. Each deployment is structured and repeatable, ensuring only stable versions reach users.
Learn more →
2. IntegrateConnect your deployed app to your products and channels.
Learn more → AI for Service, AI for Work