Kore.ai

The tools and integrations layer connects external systems over HTTP, MCP, AWS Lambda, SDK invocations, the A2A protocol, sandboxed tools, OpenAPI definitions and webhooks, with circuit breakers and retry logic, and named data sources including Salesforce, SAP, Microsoft, VMware and Zendesk reached through tool definitions rather than embedded agent logic. Breadth across protocols and system classes.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Four orchestration patterns are invoked through ABL constructs and enforced by the runtime, including parallel fan-out with barrier synchronization, asynchronous delegation that survives pod restarts, and multi-intent turns mapping to several agents; supervisor routing arbitrates handoff and delegation, a pipeline classifier short-circuits the reasoning loop for high-confidence intents, and the Google A2A protocol carries cross-runtime communication to agents hosted elsewhere. The dual-brain runtime executes agentic reasoning and deterministic flows in parallel from one definition.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Search AI is the platform's knowledge retrieval layer, providing RAG, vector search, knowledge graph traversal and reranking, surfacing context to agents during reasoning, with published guidance on model selection, chunking and context-window sizing for the retrieval pipeline. A vendor-built structure the customer configures and the platform maintains.

SourceKore.ai Agent Platform and Search AI documentation; docs.kore.ai/agent-platformread 2026-09-05

The Workflow Engine runs long-running workflows spanning minutes to days with human-in-the-loop steps and approval gates as documented primitives, guardrails are a first-class ABL construct enforced outside the model, the runtime carries multi-tier guardrails with circuit breakers and rate limits, supervisor routing is guardrail-aware so the runtime overrides an agent's chosen route when policy requires, and Arch AI surfaces its work products for human review rather than applying them silently.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Security and governance ships SSO, role-based access control, guardrails, audit logging, encryption, key management, multi-tenancy with full tenant isolation, PII detection, identity verification and compliance reporting, on an enterprise foundation carrying PCI, SOC 2, GDPR and FedRAMP certification, with right-to-erasure handling and data minimization controls. Attestation and named customer-facing controls both documented.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Reasoning-aware observability captures every reasoning step, tool call, routing decision and guardrail evaluation as structured trace data across agent delegation chains, with Deep Observability recording over 200 trace events per session and providing runtime inspection, alerting and session-level debugging across multi-agent interactions. Reconstruction of why, not reporting on what.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Memory is a formal construct in Agent Blueprint Language alongside goals, tools and guardrails, executed against a shared memory layer that coordinates state across both cognitive engines.

Kore.ai documents two stores: a system-managed Session Meta Memory, and Custom Memory Stores that are persistent, JSON Schema structured, readable and writable from code tools or workflow nodes, and assigned an access scope of user-specific persisting across all that user's sessions or session-level, with retention policies setting automatic expiry. Scope and lifetime are both stated and shared stores pass structured context between agents.

SourceKore.ai Agent Platform documentation and memory architecture writeup; docs.kore.ai/agent-platformread 2026-09-05

Two documented deployment options, cloud SaaS or self-hosted for organizations with or compliance requirements, delivering the same platform capabilities, with the infrastructure layer pluggable across Microsoft Azure, Google Cloud, AWS, on-premise installations and private VPC deployments. Customer control over where the platform runs, not a vendor architecture claim.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

The AI agent marketplace lists named prebuilt agents across banking (ATM Locator, Dispute and Fraud Handler), retail (Product Search, Order Management), HR (Absence Management, Screening Agent), IT (Incident Management, Access Management), sales (RFP Response Automation, Lead Analyzer), legal and finance (Contract Amendment Assistant) and customer service (Escalation Predictor, Contextual Email Responder), installed with one click, and the platform documentation describes importing marketplace agents with their predefined tools and workflows and editing any component afterwards.

SourceKore.ai AI agent marketplace and Create an Agent from Marketplace documentation; docs.kore.ai/agent-platform/agents/create-from-marketplaceread 2026-09-30

Agents deploy across 27 channels covering web, voice, messaging, email and enterprise APIs with consistent behavior and no per-channel rebuilding, with Microsoft Teams, ServiceNow, Slack, Webex and Zoom named. Correction to the earlier basis: the platform documentation states 27 channels, not the 40 plus figure carried from the launch release.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Every ecosystem layer is pluggable and the customer swaps providers without rewriting agents: the supported model list names OpenAI, Anthropic, NVIDIA, Cohere and LLaMA 2 among others, with open-source and custom fine-tuned models supported through the same interface.

The product documentation carries a models library where an admin adds a new model or a custom integration, a bring-your-own framework covering third-party hosted and enterprise-hosted models, per-feature model assignment, and an analytics dashboard reporting token distribution, latency and cost per model connection. Customer selection is documented, not inferred.

SourceKore.ai Agent Platform and Generative AI model documentation; docs.kore.ai/agent-platformread 2026-09-05

Agent Blueprint Language is a typed, schema-driven language compiled through DSL to intermediate representation to runtime, with definitions stored as Git-native diffable YAML reviewed in pull requests and promoted by the same CI/CD practices as application code; the tools layer exposes HTTP, MCP, SDK invocations, OpenAPI, webhooks and the A2A protocol for cross-runtime agent communication including agents hosted outside the platform, all behind a public documentation site.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Agent Evals is an automated evaluation framework providing test suites, regression tracking and quality gates that run as part of the deployment pipeline, and the environment pipeline routes every agent through a staging tier where automated suites and regression checks run before a pilot tier validates behavior on limited real traffic. Compile-time validation catches broken references before deployment. The artifact under test is the customer's own agent and the results gate promotion.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05

Agents reach external systems only through the documented tools layer, HTTP, MCP, Lambda, SDK, A2A, OpenAPI and webhooks, and reach users through 27 programmatic channels. No browser control, virtual desktop or operation of software lacking a programmatic interface appears in the platform documentation.

SourceKore.ai Agent Platform documentation; docs.kore.ai/agent-platformread 2026-09-05