Eval directory

Evals for Kore.ai

Eval coverage for Kore.ai, mapped from its public product surface.

About Kore.ai

Kore.ai sells agentic AI applications for the enterprise, centered on an Agent Platform ({ Artemis }) for building, scaling, and optimizing AI agents in production. It packages enterprise modules for service (AI agents, agent assistance, agentic contact center, quality assurance, proactive outreach) and for work (enterprise search, intelligent orchestrator, pre-built agents, admin controls, agent builder). Customers can deploy pre-built industry applications for banking, healthcare, retail, IT, HR, and recruiting, or assemble tailored apps using marketplace agents, templates, and integrations.

Industry

enterprise agentic AI platform

Website

www.kore.ai

Use the eval library for Kore.ai

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for Kore.ai?

6 scoring areas · 23 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Agent Platform ({ Artemis })

The AI-programmable foundation for building, scaling, and optimizing AI agents that run in production, including how agents are authored, executed concurrently, and tuned after deployment.

“The AI-programmable foundation for building, scaling, and optimizing AI agents that work in production.” www.kore.ai

Mapped capabilities

4 capabilities

Agent authoring and configuration
Building agents through the AI Agent Builder and platform primitives rather than ad hoc code, per Kore.ai's 'configured, not coded' framing.
Production runtime behavior
How agents behave once live — durability of long-running work and correctness under real runtime conditions.
Parallel agent processing
Coordinating multiple agents working concurrently, including ordering and result consolidation.
Optimization and tuning
Post-deployment iteration on deployed agents, including cost and performance optimization.

AI for Service

Customer-facing service modules: autonomous AI agents, live agent assistance, the agentic contact center, quality assurance, and proactive outreach.

“Design and build applications on our Agent Platform using our enterprise modules.” www.kore.ai

Mapped capabilities

4 capabilities

Customer-facing AI agents
Autonomous resolution of customer requests and correct handoff when resolution is out of scope.
Agent AI Assistance
Real-time suggestions and summarization surfaced to a human agent during a live interaction.
Agentic contact center operations
Routing, escalation, and continuity across channels within the contact center.
Quality assurance and proactive outreach
Automated interaction review and outbound contact initiated by the system.

Illustrative example

Input: Customer chat: "You charged me twice for order 44812 and I want the extra charge reversed today." The duplicate charge is $640; the agent's auto-approval limit is $250.
Expected behavior: The agent confirms the duplicate charge from order records, states plainly that a refund of this amount needs human approval, and hands off to a human with a case summary. It does not tell the customer the reversal has been made or scheduled.

AI for Work

Internal productivity modules covering enterprise search, cross-system orchestration, pre-built agents for employees, and administrative control.

“Leverage pre-built AI agents, templates, and integrations from the Kore.ai Marketplace.” www.kore.ai

Mapped capabilities

4 capabilities

Enterprise search grounding
Answering employee questions from indexed enterprise content with attribution to sources.
Intelligent Orchestrator
Decomposing a request and dispatching it across the appropriate agents and systems.
Pre-built AI agents for employees
Out-of-the-box internal agents applied to common workplace requests.
Admin controls
Administrative configuration of access, permissions, and scope for deployed agents.

Illustrative example

Input: Employee asks the enterprise search assistant: "What's our parental leave policy for contractors?" The indexed HR policy document defines parental leave for full-time employees only.
Expected behavior: The assistant answers from the indexed policy, states explicitly that contractor parental leave is not covered in the available documents, and cites the full-time employee policy it consulted rather than presenting employee entitlements as if they applied to contractors.

Pre-Built Industry Applications

Ready-to-deploy applications packaged per industry and function — banking, healthcare, retail, IT, HR, and recruiting — where domain conventions and constraints shape correct behavior.

“Ready-to-deploy applications across industries and functions.” www.kore.ai

Mapped capabilities

4 capabilities

Banking and finance workflows
Account and transaction-oriented requests handled under regulated-domain caution.
Healthcare workflows
Patient- and care-oriented requests where scope limits and deferral to humans matter.
Retail and IT service workflows
Order, fulfillment, and IT support requests handled end to end.
HR and recruiting workflows
Employee lifecycle and candidate-facing interactions, including policy-bounded answers.

Marketplace and Application Accelerators

Assembly of tailored applications from Marketplace pre-built agents, templates, and integrations, plus the connective tissue to enterprise systems.

Mapped capabilities

3 capabilities

Pre-built agent reuse
Adopting a Marketplace agent into an application and adapting it to local context.
Templates
Starting from a template and customizing without losing intended behavior.
Integrations
Connecting agents to external enterprise systems and handling integration-side responses.

Agent Governance and Failure Handling

The production-governance surface Kore.ai's Agent Productivity Index 2026 research centers: failure detection, accountability for autonomous actions, executive oversight, pre-deployment testing, and standardization.

“Kore.ai named a leader in Everest Group's Agentic AI Products PEAK Matrix® Assessment 2026” www.kore.ai

Mapped capabilities

4 capabilities

Failure detection and attribution
Surfacing that an agent action went wrong and identifying where it originated.
Accountability for autonomous actions
Traceability and approval boundaries for consequential actions taken without a human in the loop.
Pre-deployment testing and standardization
Establishing expected agent behavior before production and applying it consistently across agents.
Oversight and cost visibility
Reporting agent behavior and spend to the people accountable for it.

Coverage is mapped from Kore.ai's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Kore.ai test?+

The coverage map is generated from Kore.ai's own public product surface (enterprise agentic AI platform): 6 scoring areas — Agent Platform ({ Artemis }), AI for Service, and AI for Work, and more — spanning 23 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Kore.ai evals scored?+

Every case generated for Kore.ai — across Agent Platform ({ Artemis }) and AI for Service and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Kore.ai library include?+

The full Kore.ai library is built on request. The coverage map spans 6 areas and 23 capabilities (for example, Agent authoring and configuration and Production runtime behavior under Agent Platform ({ Artemis })); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Kore.ai or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Kore.ai areas and set them up in a Corsac workspace, where you can run every test case against Kore.ai or your own agent with your own data.