Delight.ai

Named systems on both sides of the agent, with retrieval and handoff documented separately. Delight.ai publishes integrations with Zendesk, Salesforce and Freshworks for live agent handoff, with dedicated product-release notes for Salesforce supported as a knowledge source and for the Freshworks integration, plus third-party knowledge base integration so the agent can surface the right information in real time.

Agents are described as taking action rather than only answering, and Trust OS logs every tool call, which is direct evidence that tool invocation is a first-class mechanism rather than a claim. On the channel side the agent reaches customers through in-app chat, web, email, SMS, WhatsApp and social, all built on Sendbird's communications platform, and a dedicated AI Agent Integrations page carries the connector set.

SourceDelight.ai integrations page, AI agent platform page and Salesforce and Freshworks product releasesread 2026-09-05

Multi-step ownership of work that outlives a single conversation. Agent Steward is published as taking full ownership of complex, long-running customer issues end to end rather than answering and closing, and the vendor frames the distinction against chatbots explicitly: agents understand intent, hold context, take action and adapt, resolving complex issues and escalating only when needed.

Behavior is orchestrated from a customer-authored actionbook holding rules such as refund thresholds and escalation logic, which Openbook renders as a visual flow through complex branching. Above the individual case, Zero-Touch Improvement runs a closed loop in which observability catches an issue, control tests the fix and oversight holds it to policy, described as a graduated system evolving from human-guided to autonomous.

SourceDelight.ai Agent Steward, Trust OS and Omnipresent pagesread 2026-09-05

A maintained corpus the customer owns and connects, kept separate from the memory layer. Delight.ai supports third-party knowledge base integration so the agent can access and surface the correct information in real time, with Salesforce documented as a supported knowledge source in its own product release and Freshworks and Zendesk connected alongside.

Building an agent is described as transferring knowledge to it, like onboarding a team member, through the supervisor dashboard, and the actionbook holds the authored rules the agent answers within. Grounding quality is measured rather than assumed: Trust OS detects factual errors and off-policy responses automatically, and knowledge-side accuracy is what the review-and-grade surface scores.

SourceDelight.ai AI agent platform and builder pages, and the Salesforce knowledge-source releaseread 2026-09-05

An operator control surface with a hard stop. Trust OS publishes kill switch and escalation controls: pause, redirect or kill any AI flow instantly, with one-click escalation to a human agent and role-based access control so the right people own the right agents. Policy itself is customer-authored through the Actionbook Editor, which gives operations teams direct authorship over agent behavior with no code and no engineering ticket, editing refund thresholds, escalation logic and tone in real time.

Safety runs automatically alongside: hallucination and unsafe content detection flags factual errors, off-policy responses and safety violations before customers see them, and a review-and-grade surface puts humans in the position of evaluators, scoring responses and flagging outliers against the customer's own standards. Handoff to a person is wired into the support stack through Zendesk, Salesforce and Freshworks integrations. Oversight comes through authored policy, automatic detection and interruption rather than a sign off on each action before it is sent.

SourceDelight.ai Trust OS and AI agent platform pagesread 2026-09-05

Attestations and customer-facing controls are both documented, with a live trust portal. SOC 2 Type 2, HIPAA, GDPR and CCPA are named on the security page, with ISO 27001 and HIPAA/HITECH carried on the platform page, and a trust center is published at trust.sendbird.com, a live control portal rather than a badge.

Granular role-based access controls and full audit logs are named as customer-facing platform controls, alongside flexible data retention settings the customer aligns to their own policy, end-to-end encryption and TLS in transit, PII protection and data masking. RBAC appears again on the governance side, scoping which people own which agents.

Internal posture is separately described as FIDO2 multi-factor authentication with device trust, automated detection and response, automated compliance and a DevSecOps development approach, and a vulnerability disclosure address is published.

SourceDelight.ai security page, AI agent platform page and Trust OS pageread 2026-09-05

Per-decision tracing published as a named product rather than inferred from dashboards. Trust OS states that everything is logged and easy to trace, and Openbook is documented as logging every conversation, every decision and every tool call for review, with the vendor's own phrasing that there are no black boxes. Openbook also renders the agent's actionbook as a visual flow so an operator can navigate the logic behind a decision rather than read markup.

Around the trace sit diagnostic layers: Low CSAT Analysis sorts every conversation scored one or two into one of five failure types before anyone opens it, and Thinking Messages surface a per-tool-call status during the run so the reasoning is visible in flight as well as after. A dedicated Analytics product tracks conversations, common topics and ROI. This records the vendor's own agent.

SourceDelight.ai Trust OS and Analytics pagesread 2026-09-05

A memory layer sold as its own product. The Agent Memory Platform is published as delight.ai's intelligence layer giving agents a living memory and a business brain, connecting what the agent remembers about each customer with what the business is trying to achieve.

Persistence is the product's founding claim rather than a side effect: agents absorb signals from every conversation and action to form an evolving understanding of the customer, explicitly positioned beyond static CRM data, and context is maintained as a customer moves between chat, SMS, email and voice so nobody repeats themselves.

The platform page states it directly as an agent that remembers every interaction, and continuity across separate conversations, not just within one, is what the vendor sells against conventional chatbots that lose history the moment a session ends.

SourceDelight.ai Agent Memory Platform and AI agent platform pagesread 2026-09-05

The buyer controls placement, stated in those words. The security page publishes data privacy and residency options under the heading that a customer maintains full control over where their data resides, how it is stored and who can access it, in order to meet regulatory and internal data governance requirements, and names regional hosting to support data residency needs.

The platform page carries the same commitment from the product side, listing region-specific storage among its privacy-first data governance properties alongside PII protection and data masking. Retention is customer-set rather than fixed, with flexible data retention settings aligned to the customer's own policy, and data ownership and processing are described as transparent. No on-premise or self-hosted deployment is offered.

SourceDelight.ai security and AI agent platform pagesread 2026-09-05

Named capability modules a buyer switches on, but not a catalog they browse. Delight.ai publishes a capabilities set each with its own page - Agent Steward for autonomous ownership of long-running issues, Agent Memory Platform, For You Conversations, Omnipresent, Voice AI Agent, Trust OS, Analytics, Desk and AI Agent Integrations - and a buyer adopts these as units rather than assembling them. Industry pages exist for retail, travel, healthcare, financial services, logistics and more.

No template gallery, agent library or marketplace a customer browses and puts into service is documented. The documented route to an agent is construction, not adoption: the builder is described as onboarding an agent like a new team member by transferring knowledge and providing workflows through a supervisor dashboard, and behavior is authored in an actionbook.

SourceDelight.ai AI agent builder, capabilities index and industry pagesread 2026-09-05

Breadth plus a documented initiating path, both first-party. The channels are in-app chat, website, email, SMS, WhatsApp, social messaging and voice, all carried on Sendbird's communications platform, with the agent meeting customers wherever they are and context following them between channels rather than restarting.

Initiation is not only inbound: Omnipresence is published as always on, always aware and always personal, with agents anticipating needs and proactively starting conversations before an issue forms, and re-engaging when a customer goes quiet or switches channel. Voice 2.0 adds advanced multilingual real-time outbound calling, which is an initiating channel rather than a receiving one.

SourceDelight.ai Omnipresent, Voice AI Agent and AI agent platform pagesread 2026-09-05

No model selection, routing configuration or bring-your-own-key path appears on the AI agent platform page, the agent builder description, Trust OS, the security page or the capabilities index; all are silent on which models run the agent or who chooses them. The product is sold as a managed outcome rather than a configurable runtime: resolution rates, a governance layer that detects hallucinations, and zero-touch improvement that tunes the agent autonomously.

SourceDelight.ai AI agent platform, Trust OS and security pagesread 2026-09-05

A developer surface that predates the agent product and carries it. Sendbird publishes public AI agent documentation at sendbird.com/docs/ai-agent, linked from delight.ai's own developer menu alongside a public community forum and a server status page, plus a GitHub organization at github.com/sendbird.

The underlying communications platform ships cross-platform SDKs for iOS, Android, JavaScript, React Native and Unity with documented chat, voice and video APIs, which is how agents are embedded natively in a customer's own app rather than bolted on as a widget. On the agent side, AI Agent Integrations connect the agent outward to Zendesk, Salesforce and Freshworks, and Trust OS exposes the actionbook as an editable artifact operations teams author directly. No MCP server or agent manifest surface is documented.

SourceSendbird AI agent documentation, delight.ai developer menu and the Sendbird communications API platform; sendbird.com/docs/ai-agent/guide/v1/overviewread 2026-09-05

A staged release pipeline with hard comparable results. Trust OS publishes the principle outright, that every change is tested before it touches a customer, with staging, real-conversation test suites, gradual rollout and A/B data so a deploy comes with evidence rather than faith. Staging environments are described as a production mirror carrying real integrations and policies.

Test Upgrade builds test cases from real conversation IDs, organizes them with labels, shares sets across environments and returns consistent pass/fail results. Gradual rollout ships to ten percent first and A/B testing runs two versions head to head for hard data on resolution rate and CSAT before committing.

Separately the platform page describes simulated environments for testing and evaluating an agent before deployment, and Trust OS carries a review-and-grade surface where every response gets a quality score.

SourceDelight.ai Trust OS and AI agent platform pagesread 2026-09-05

No browser control, hosted session, virtual desktop or interface automation appears anywhere across the platform, capabilities, Trust OS or security pages. The route out is stated positively and it is programmatic: the agent acts through integrations with Zendesk, Salesforce and Freshworks, through tool calls that Trust OS logs individually, and through the communications APIs and SDKs that embed it in the customer's own app. Its surface to the world is conversational - in-app chat, web, email, SMS, WhatsApp, social and voice - rather than a screen it operates.

SourceDelight.ai AI agent platform and Trust OS pagesread 2026-09-05