LivePerson

Open platform is published as a Conversational Cloud pillar: LivePerson's platform is brought together with the customer's existing CRM, marketing and data platforms to personalize and streamline interactions and break down technology silos, with a dedicated integrations catalog page and a partner marketplace of certified integrations alongside. The developer center documents Data APIs for moving conversational data in and out programmatically, so the connection is a documented interface rather than a fixed connector list.

Sourceliveperson.com/products/integrationsread 2026-09-05

The platform orchestrates a mixed workforce and the channels it works across, at enterprise scale: human agents, intelligent automations and conversational AI are balanced within one conversation across voice and messaging, intent identification drives what happens next, Voice AI automates the voice channel end to end with LLM-powered bots, and the Voice to Digital solution routes inbound calls from an existing CCaaS estate into digital threads.

An agent and supervisor workspace carries handoff, review and escalation. Flow building and multi step branching are not described on the platform pages.

Sourceliveperson.com/products/conversational-cloudread 2026-09-05

An intent layer is documented and a retrieval structure is not.

LivePerson publishes natural language understanding and natural language processing as the basis for human-like interaction, positions the Conversational Cloud as an AI platform that identifies customer intent, and states that nearly a billion monthly conversational interactions make its intent models more accurate than competitors; Syntrix independently confirms intent models exist by stress-testing intent accuracy.

No index, embeddings layer, retrieval architecture, refresh behavior or source level citation is documented, and KnowledgeAI is not described, so grounding against a maintained corpus is not shown.

Sourceliveperson.com/conversational-airead 2026-09-05

An explicit pre-send approval surface: the Conversational Cloud page states that agents can stay in the loop to review, edit and approve AI responses before they are sent. That is a vendor-owned gate ahead of the action rather than routing after it.

Around it, the platform is built for human agents, automations and AI to share one conversation with handoff between them, the agent and supervisor experience adds real-time context and supervisor visibility, and Syntrix stress-tests flows, handoffs and guardrails in a pre-production environment, which independently confirms those controls exist to be tested.

Sourceliveperson.com/products/conversational-cloudread 2026-09-05

Compliance and access are both documented. A live Conveyor-powered trust center at trust.liveperson.com carries SOC 2 Type II and ISO/IEC 27001:2022 with certifications downloadable, alongside answer libraries spanning access management, application and data security, artificial intelligence, cloud security, continuity, incident management, privacy and security governance.

The security page adds SSAE 18 SOC2+HIPAA reporting audited yearly to AICPA standards, PCI DSS 4.0, GDPR, Sarbanes-Oxley, HIPAA via executed BAAs, and HITRUST Common Security Framework audit, plus a formal Vulnerability Disclosure Policy and Program and roughly 30 penetration tests a year per the developer documentation. For sign in, the Conversational Cloud page names integration with the brand's own single sign-on experience, alongside available masking and encryption of data at rest, firewalls, ACLs, load balancers and DDoS protection. The security page describes the ISMS as aligned with ISO 27001 while the trust center lists ISO/IEC 27001:2022 as held, and the trust center is the stronger statement.

Sourcetrust.liveperson.comread 2026-09-05

Analytics and supervisor visibility are documented; run reconstruction is not. Conversational Intelligence and Insights ships as its own product, giving access to and analysis of all conversational data across voice and messaging to understand what customers want and to improve contact center performance, and the agent and supervisor experience is published as increasing supervisor visibility with real-time context.

No step level decision trace, tool call log or retrieval record for an individual automated interaction is documented, so a buyer can measure outcomes but not reconstruct how one conversation was resolved. Syntrix provides transcript level traceability, but only over simulated conversations before production.

Sourceliveperson.com/products/conversational-intelligenceread 2026-09-05

What persists is not stated. Agents work in the Conversation Manager workspace, and the Conversational Cloud page describes giving them context in real time without saying what is retained or for how long; the Conversational AI, Syntrix and security pages say nothing either. Conversational Intelligence gives access to and analysis of all conversational data across voice and messaging, but that is an analytics corpus rather than state an agent carries, and a store is not a memory layer whoever writes it. 2026-09-05

No residency position appears on the product or security pages. The security page describes yearly data center audits and N+1 power across all data centers but names no region and offers no selection, and the Conversational Cloud, Syntrix and Conversational AI product pages say nothing on residency. LivePerson sells to HSBC, Zurich Insurance and other regulated European brands, but where their data sits is not stated there. 2026-09-05

Adoptable assets exist but are assembled or delivered rather than browsed. LivePerson publishes an AI Chatbot Toolset as a named bundle for building bots, Voice AI as a packaged automation for the voice channel, and a partner Marketplace of solutions delivered by certified partners, which is an adoption route the buyer can take. No vendor supplied catalog of prebuilt agents or templates that the customer browses and puts into service is documented; the routes that do exist run through partners or through building on the toolset.

Sourceliveperson.com/products/ai-chatbotsread 2026-09-05

Breadth across voice and messaging with the agent both receiving and initiating.

The Conversational Cloud page publishes communication channels as a platform pillar covering SMS, WhatsApp, voice and other key channels, with a dedicated Messaging Channels product page, multi-channel deployment and multi-language support across geographies; Voice AI automates the voice channel with LLM-powered bots, and the Voice to Digital solution moves inbound calls from a CCaaS estate into digital threads.

Initiation is documented on the financial services solution page, which describes proactively engaging consumers. The acquirer's completion release positions the combined company as omnichannel conversational AI.

Sourceliveperson.com/products/messaging-channelsread 2026-09-05

No model selection, routing control or bring your own model option is exposed to the customer, and no provider is named. LivePerson describes delivering outcomes with the power of state-of-the-art large language models and ships LLM-powered voice bots, so external models are plainly in use, but which ones and who chooses are undisclosed on the product surface. The trust center carries an Artificial Intelligence answer library of over two hundred entries.

Sourceliveperson.com/products/generative-airead 2026-09-05

A public developer center at developers.liveperson.com, linked from the site footer, is published as a digital toolbox for building on the Conversational Cloud, with an All Documents index, a Data APIs section and security documentation covering consumer authentication. A developer can read the surface and authenticate against it without a sales conversation. No Model Context Protocol server is documented.

Sourcedevelopers.liveperson.com/all-documents.htmlread 2026-09-05

Syntrix ships as a named evaluation product. Before launch it simulates real customer interactions in a pre-production environment, runs multiple simulated conversations simultaneously to validate agent performance across edge cases, stress-tests intent accuracy, and tests guardrails and compliance handling, against synthetic personas built to behavioral profiles.

The result is readable and comparable: the published cycle is Test, Evaluate, Verify and Validate, with scorecards and analytics producing actionable insights, a summary view of outcomes across every simulated conversation, drill-down into individual transcripts with full traceability, and performance segmented by agent. It runs continuously after launch as well as before it, and the vendor reports a 60 percent reduction in bot testing cycles.

Sourceliveperson.com/syntrixread 2026-09-05

The route out is programmatic, and LivePerson states it positively: the open platform connects to the customer's CRM, marketing and data systems, the integrations catalog and Data APIs carry the traffic, and automation reaches customers over voice and messaging channels rather than by operating an interface. No browser control, hosted session, virtual desktop or autonomous computer use is documented. Cobrowse, where a human agent views and guides a customer's session, is not documented either, and it would not be autonomous browser control in any case.

Sourceliveperson.com/products/conversational-cloudread 2026-09-05