Decagon vs Kore.ai (2026)

Connectors span CRM, helpdesk and ticketing (Salesforce, Intercom, Zendesk), knowledge bases (Confluence, Contentful, Kustomer), and CPaaS and telephony (Amazon Connect, RingCentral, SIP trunking), plus custom endpoints and MCP connectivity for any other system. Agents both retrieve data and take action.

Sourcedecagon.ai/product/integrationsread 2026-09-05

The tools and integrations layer connects outside systems over HTTP, MCP, AWS Lambda, SDK invocations, the A2A protocol, sandboxed tools, OpenAPI definitions and webhooks, with circuit breakers and retry logic. Data sources such as Salesforce, SAP, Microsoft, VMware and Zendesk are reached through tool definitions, so none of them is built into agent logic.

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

Agent Operating Procedures turn natural language inbound and outbound workflows into executable agent logic that pulls live data and triggers actions across connected systems. Versions are tracked in Git, and rollout is staged across agent versions.

Sourcedecagon.ai/product/aopread 2026-09-05

Four orchestration patterns are invoked through ABL constructs and enforced by the runtime. They include parallel fan out with barrier synchronization, asynchronous delegation that survives pod restarts, and turns carrying several intents that map to several agents.

Supervisor routing arbitrates handoff and delegation, a pipeline classifier skips the reasoning loop for intents it recognizes with high confidence, and the Google A2A protocol carries communication to agents hosted on other runtimes. The runtime's two engines execute agentic reasoning and deterministic flows in parallel from one definition.

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

Chat, voice and email run on one intelligence layer. Inbound calls route over SIP and CPaaS platforms, email comes in through Zendesk and Intercom, live chat escalates through Zendesk Sunshine and Salesforce, and outbound and proactive agents are supported as a core feature.

Sourcedecagon.ai/product/overviewread 2026-09-05

Agents deploy across 27 channels covering web, voice, messaging, email and enterprise APIs, among them Microsoft Teams, ServiceNow, Slack, Webex and Zoom, with consistent behavior and no rebuilding for each channel.

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

Decagon maintains the agent's own knowledge base, filled by sync from customer systems including Confluence, Contentful and Kustomer. Suggestions keeps it current by finding gaps from conversations, ranking their impact on coverage and issuing monthly content updates. Traces name the knowledge articles the agent referenced. The knowledge base persists between conversations.

Sourcedecagon.ai/product/suggestionsread 2026-09-05

Search AI is the platform's knowledge retrieval layer, providing RAG, vector search, knowledge graph traversal and reranking to surface context to agents as they reason. The customer configures it and the platform maintains it, and the retrieval pipeline comes with guidance on model selection, chunking and context window sizing.

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

Memory is a named part of the agent engine. It holds customer context across sessions and channels alongside the customer's systems of record, with stated data portability and control, so a conversation started in chat continues by voice without repetition. Decagon does not publish how long memory is kept or how to delete it for one customer.

Sourcedecagon.ai/product/overviewread 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. Session Meta Memory is managed by the system.

Custom Memory Stores are persistent and structured with JSON Schema, and each takes a scope: one user across all of that user's sessions, a single session that clears when it ends, or the whole application. A retention policy sets each store to last a session, a day, a week or a month, and tools read, write and delete store content at run time. Shared stores pass structured context between agents.

Sourcedocs.kore.ai agent platform memory stores guideread 2026-10-25

Guardrails can be configured for brand voice, escalation and hallucination. A supervisor model revises ungrounded responses before they are sent, bad actor detection escalates to a human, and Watchtower flags conversations for triage against criteria the customer sets. Oversight is automated gating and quality review after the fact, with no customer facing approval queue that holds an agent action until someone signs off.

Sourcedecagon.ai/product/aopread 2026-09-05

The Workflow Engine runs workflows lasting minutes to days, with built in steps where a person reviews the work and approval gates that hold it until someone signs off. Guardrails are a core ABL construct enforced outside the model, the runtime carries guardrails in several tiers with circuit breakers and rate limits, and supervisor routing checks those guardrails, so the runtime overrides an agent's chosen route when policy requires. Arch AI hands its work products to a person to review and never applies them silently.

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

Access controls include role based access control, SSO with Okta and Microsoft Entra, tamper protected audit logs and short lived scoped JWTs for agent access. Data is encrypted with AES-256 at rest and TLS 1.2+ in transit, DLP based redaction removes PII, and model providers are held to zero-day retention. Decagon shows SOC 2, ISO 27001, PCI DSS, HIPAA, CCPA and Data Privacy Framework marks and runs a live trust portal.

SourceDecagon security page and trust.decagon.ai; decagon.ai/securityread 2026-09-05

Security and governance include SSO, role based access control, guardrails, audit logging, encryption, key management, multitenancy with full tenant isolation, PII detection, identity verification and compliance reporting, on an enterprise foundation carrying PCI, SOC 2, GDPR and FedRAMP certification, with handling for the right to erasure and data minimization controls.

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

Traces explain how and why the agent decided as it did at any point in a conversation, covering model calls, workflow triggers, the knowledge articles referenced, and runtime and latency figures. Watchtower reviews every conversation against criteria the customer defines and flags outliers for triage.

Sourcedecagon.ai/product/overviewread 2026-09-05

Observability follows the agent's reasoning, capturing every reasoning step, tool call, routing decision and guardrail evaluation as structured trace data across chains of delegated agents. Deep Observability records over 200 trace events per session and provides runtime inspection, alerting and debugging of each session across interactions between agents.

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

The platform runs on multi region infrastructure with autoscaling and automatic failover for resilience. There is no customer selectable region, VPC, single tenant or on premises option, so where data sits depends on Decagon's architecture, not the customer's choice. The product is delivered only as SaaS hosted by Decagon.

Sourcedecagon.ai/securityread 2026-09-05

Kore.ai runs as cloud SaaS or self hosted for organizations with data residency or compliance requirements, with the same platform capabilities either way. The infrastructure layer plugs into Microsoft Azure, Google Cloud, AWS, on premises installations and private VPC deployments, so the customer chooses where the platform runs.

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

Customers build agents by writing Agent Operating Procedures in natural language, or Duet generates them from the customer's own conversation transcripts, applying Decagon's deployment learnings behind the scenes. The seven industry pages are positioning and case studies, not solutions to adopt. There is no template gallery, named prebuilt agent set, blueprint library or industry pack.

Sourcedecagon.ai/product/aopread 2026-09-05

The AI agent marketplace offers 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. Marketplace agents import with their predefined tools and workflows, and any component can be edited 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

The agent engine runs on a network of frontier and specialized models, with OpenAI and Anthropic named as providers held to zero-day retention, and model redundancy is a platform property. Routing happens inside Decagon, and customers and admins cannot select the model that powers the product.

Sourcedecagon.ai/product/overviewread 2026-09-05

Every layer of the ecosystem is pluggable, so the customer swaps providers without rewriting agents. Supported models include OpenAI, Anthropic, NVIDIA, Cohere and LLaMA 2 among others, and open source and custom fine tuned models work through the same interface. A models library lets an admin add a new model or a custom integration, models hosted by a third party or by the enterprise itself can be brought in, each feature can be assigned its own model, and an analytics dashboard reports token distribution, latency and cost per model connection.

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

A REST API for the platform sits at api.decagon.ai, authenticated with a customer issued key passed as a bearer token, and rate limited. An API key can be generated from the Developer page of the Decagon dashboard. The documentation at docs.decagon.ai sits behind a customer login, nothing on the public site links to it, and the public Integrations page describes only connections into the customer's own systems.

SourceDecagon developer documentation at docs.decagon.airead 2026-09-05

Agent Blueprint Language is a typed language driven by schemas, compiled from DSL to intermediate representation to runtime, with definitions stored in Git as 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 communication with agents on other runtimes, including agents hosted outside the platform.

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

Testing & QA and Experiments are customer facing products. Agents are validated before launch with end to end simulated conversations and unit tests. Live traffic is then routed across agent versions to A/B test changes and measure their effect on CSAT, deflection rate and other metrics at scale, with readable results the customer can compare.

Sourcedecagon.ai/product/overviewread 2026-09-05

Agent Evals is an automated evaluation framework for the customer's own agents, with test suites, regression tracking and quality gates that run as part of the deployment pipeline, and the results decide whether an agent is promoted. 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. Validation at compile time catches broken references before deployment.

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

The agent acts only through API and MCP connectors, CRM and ticketing integrations, CPaaS platforms and SIP telephony. It does not control a browser or virtual desktop or operate software that lacks a programmatic interface.

Sourcedecagon.ai/product/integrationsread 2026-09-05

Agents reach outside systems only through the tools layer (HTTP, MCP, Lambda, SDK, A2A, OpenAPI and webhooks) and reach users through 27 programmatic channels. They do not drive a browser, a virtual desktop or software that lacks a programmatic interface.

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