Enterprise conversational AI platforms compared (2026)

AI customer service platform that runs AI agents across web chat, messaging, email and voice channels.

Vendor: Ada Support Inc.

Models named: OpenAI, Anthropic, Microsoft Azure OpenAI, Amazon AWS, Cohere, Groq, Baseten

Pricing: Quote based; charged per AI agent conversation, with a per resolution option for some enterprise customers.

Notes: PCI DSS is unknown: the trust center lists a PCI Attestation of Compliance without date or scope, while the Customer Terms of Use (updated August 2026) state the Services are not PCI DSS compliant; ISO 27001 is not listed (Ada follows CIS Controls v8). HIPAA rests on a HIPAA report in the trust center and the HIPAA Business Associate Agreement referenced in the customer terms; payments are unknown because the voice page lists payment processing while the same terms forbid requesting card or banking details. Providers are taken from the subprocessor list (AI models rows), which does not say which model serves which task.

Agentic AI platform for automating IT, HR and customer service requests, now part of Automation Anywhere.

Vendor: Automation Anywhere (acquired Aisera, announced 4 November 2025)

Models named: Aisera (proprietary LLM), Anthropic, Meta, Amazon

Pricing: Custom quote via demo or RFP request

Notes: Aisera's own security page (archived October 2025, now redirecting to Automation Anywhere's security page) stated ISO/IEC 27001 certification, a yearly SOC 2 Type II audit and that Aisera signs customer HIPAA BAAs; Automation Anywhere's own certifications are not counted for Aisera. The docs mention a cloud or on premises administrative setup and a Remote Executor for reaching on premises systems, but no page confirms a self hosted Aisera product, so on_premises stays unknown. No price list is published on aisera.com, yet third party sites cite an Azure Marketplace listing with list prices, so pricing.public is unknown; Bedrock models (Anthropic, Meta, Amazon) are named on the AWS partner page alongside Aisera's proprietary LLM.

AWS service for building voice and text conversational interfaces, often used with Amazon Connect for contact center self service.

Vendor: Amazon Web Services

Models named: Amazon (built in ASR and NLU), Anthropic (Claude via Amazon Bedrock)

Pricing: Pay per request: $0.004 per speech request and $0.00075 per text request, with separate streaming conversation pricing; generative features are billed through Amazon Bedrock.

Notes: Generative features (QnAIntent, assisted NLU, assisted slot resolution) let the builder choose a provider and foundation model in Amazon Bedrock; the only provider AWS's Lex docs name is Anthropic Claude 3. AWS says the services in its ISO/IEC 42001 scope are listed on a certificate available only through AWS Artifact, so ISO 42001 for Lex is unknown. Agentic workflows, handover with an escalation summary and agentless outbound dialing come from Amazon Connect (agentic self service AI agents and outbound campaigns), not from Lex alone.

Conversational and agentic AI platform for enterprise voice and digital agents, with about half of its customers in healthcare.

Vendor: Avaamo, Inc.

Models named: Microsoft, Amazon Web Services

Pricing: Not publicly documented

Notes: The SafeBase trust center lists a SOC 2 Type 2 report, an ISO/IEC 27001:2022 document and a PCI DSS certificate with AOC, the security page states Avaamo is HIPAA compliant, and the trust center also lists HITRUST, FedRAMP 20x and CSA STAR. The current site focuses on voice agents and product documentation sits behind a login; chat evidence is a December 2021 announcement and the on premises option comes from the partner program page, so both may be dated, and outbound rests on agents that send patient reminders and notifications. The LLaMB framework is described as supporting any LLM of the customer's choice, with Microsoft and AWS LLM services named in February 2024 press coverage. Payments is not counted: the source mentions payments as a use case without describing a payment flow inside the conversation.

Platform for AI phone agents handling inbound and outbound calls, running on speech and language models hosted by Bland.

Vendor: Bland AI, Inc.

Pricing: Per minute talk and transfer time ($0.12 to $0.14/min talk time) with a monthly platform fee on the Build tier; custom Enterprise contracts. LLM, STT and TTS included in the per minute rate.

Notes: Arabic appears in Bland docs only as a live translation language for warm transfers (https://docs.bland.ai/changelog/09_02_2025), not in the native STT or TTS lists found; the trust center lists ISO 27001:2022, SOC 2 Type II, a PCI DSS 4.0.1 AOC and FedRAMP 20x Class A. On premises, VPC and air gapped deployment, a BAA, warm transfers and guardrails are Enterprise tier options; 'proprietary' reflects Bland's statement that its voice, language and speech models are built in house, without naming base models. Payments rests on a Bland article describing phone payment flows where the agent handles the conversation and an isolated IVR captures the card digits.

Enterprise platform for conversational AI, digital humans, agentic AI, custom LLMs and agentic payments, deployable in the cloud or on premises.

Vendor: Blits.ai

Models named: OpenAI, Azure OpenAI, Anthropic, Google, Mistral, Meta, Alibaba Qwen, DeepSeek

Pricing: Annual plans with a monthly interaction allowance and named seats, prices on request

Notes: Blits.ai publishes this dataset. Its row is sourced from its own public pages, the same standard applied to every other vendor.

Conversational AI platform combining intent based NLU with LLMs for chat and voice customer service, aimed at regulated industries.

Vendor: Boost AI AS

Models named: OpenAI

Pricing: Not publicly documented

Notes: Automation Anywhere signed a definitive agreement on 7 October 2026 to acquire boost.ai, expected to close in Q4 2026. ISO 27001 is stated as certified since May 2021 for the whole company, and the SOC 2 Type II audit announced in April 2026 covered a three month period from September 2025. On premises, OpenAI and the option to use the customer's own LLM API key come from contract terms versioned February 2024 and may be outdated; the acquisition release cites only "HIPAA alignment", and payments in a partnership with Eckoh rely on Eckoh's own PCI DSS accreditation, not boost.ai's.

AI agent platform for automating contact center customer service across voice and digital channels.

Vendor: NiCE (Cognigy)

Models named: OpenAI, Azure OpenAI, Anthropic, Amazon Bedrock, Google Gemini, Mistral AI, Aleph Alpha

Pricing: Not publicly documented

Notes: Cognigy docs state on premises installations are no longer offered to new customers, while existing installations keep running and receive updates. The trust center states ISO 27001, ISO 42001 and SOC 2 Type II certifications under a unified NiCE governance framework, while the independently assessed PCI DSS v4.0.1 status (AOC on the trust center) covers only deployments on CXone instances and excludes Cognigy Live Agent. Cognigy moved from Leader (2025) to Visionary (2026) in the Gartner MQ.

Blits.ai vs Cognigy, in detail →

Conversational AI platform for customer service agents across chat, email, voice and SMS.

Vendor: Decagon AI, Inc.

Models named: OpenAI, Anthropic

Pricing: Per conversation or per resolution pricing; no rates published.

Notes: Trust center lists an ISO 27001:2022 certificate (issued 22 January 2026), SOC 2 Type II reports and a PCI DSS 4.0.1 Attestation of Compliance (SAQ D for Service Providers, a self assessment); HIPAA appears only as a badge without status and Decagon pages speak of 'HIPAA options' with no BAA found, so HIPAA is unknown. Voice claims 70+ languages on Decagon's own speech model (Chord), but no language list naming Arabic was found. Payments rests on one sentence in the Dialogues 2026 announcement that agents collect payments, without technical documentation; agent logic is written as natural language AOPs and no visual builder was found. Payments is not counted: the source mentions payments as a use case without describing a payment flow inside the conversation.

Agentic AI platform for building and orchestrating enterprise AI agents over chat and voice, deployable in cloud, on premises and edge environments.

Vendor: Druid AI

Models named: OpenAI, Azure OpenAI, Anthropic, Google Vertex AI, Mistral AI, AWS Bedrock, Meta Llama, Mesolitica, GPT OSS, Druid Becus

Pricing: Custom quote based on use case, scale and deployment requirements

Notes: Druid lists SOC 2 Type II, ISO 27001:2022 and HIPAA as standards its platform is designed to support (and as readiness on its security page), without a clear certification claim, so those fields stay unknown; air_gapped reflects documentation that edge deployments may run in disconnected networks, subject to customer security policies. Payments in conversation are documented through a Netopia integration that sends a payment link, with the transaction completed on the Netopia gateway. Headquarters is reported as Bucharest, Romania (CB Insights) and as New York (Bloor Research); Gartner moved Druid from Challenger in 2025 to Niche Player in 2026.

Platform for building, testing, deploying and monitoring voice, chat and email AI agents on top of ElevenLabs speech models.

Vendor: ElevenLabs

Models named: OpenAI, Anthropic, Google (Gemini), ElevenLabs hosted open models (DeepSeek, GLM, Qwen)

Pricing: Subscription plans from free to Business with included call minutes and concurrency; extra minutes $0.08 each, text messages $0.003 each, LLM usage billed separately; Enterprise custom.

Notes: The digital human channel relies on a documented integration with HeyGen LiveAvatar, not a native ElevenLabs avatar, and private deployment runs ElevenAgents in the customer's AWS or GCP account or own hardware (Azure announced for H2 2026). ElevenLabs describes London as its headquarters, while its contracting entity is Eleven Labs Inc. and some directories list New York. The compliance portal publishes ISO 27001 and ISO/IEC 42001:2023 certificates (the latter issued to Eleven Labs, Inc. for its production platform) and a 2026 SOC 2 Type 2 plus HIPAA report, while in call payments use native DTMF capture with redaction and need the payment processor connected through tools or integrations such as Stripe.

Customer service AI agent that resolves queries across chat, email, messaging and phone, natively in the Intercom helpdesk or on top of other helpdesks.

Vendor: Fin (formerly Intercom), a Salesforce company

Models named: Fin (Apex, proprietary), OpenAI, Anthropic, Google

Pricing: $0.99 per Fin outcome (resolution, procedure handoff, disqualification or self serve routing) and $9.99 per qualification; Fin is included in all Intercom helpdesk plans.

Notes: Intercom renamed itself Fin in May 2026 (keeping Intercom as the helpdesk product name), and Salesforce completed its acquisition of Fin on 10 September 2026. Final answers are written by the in house Apex model, with automatic switching across OpenAI, Anthropic and Google models for resilience; the trust center publishes a SOC 2 Type 2 report, ISO 27001 and ISO 42001 certificates and a HIPAA report, while Intercom's terms state the Services are not PCI DSS compliant. Fin can issue refunds through Stripe or Shopify connectors, but taking a customer payment in the conversation is not documented.

AI support agent for voice, chat and email that runs on top of existing helpdesks, aimed at fintech, banking, insurance and healthcare.

Vendor: Fini Inc.

Models named: OpenAI, Anthropic

Pricing: Per resolution: $0.89 (Growth), $0.69 (Scale), $0.49 (Enterprise), no seat fees; voice billed per answered call ($0.89 down to $0.35 by volume) or per minute.

Notes: Fini offers dedicated single tenant deployment with private VPC isolation on Enterprise, operated by Fini rather than in the customer's own cloud account, and scopes an in datacenter setup only when a regulation requires it, so private cloud and on premises are unknown. Fini states it holds SOC 2 Type II and ISO 27001 (both listed with gated documents in its trust center at security.usefini.com) and also claims PCI DSS Level 1 on marketing pages, but PCI DSS is absent from the trust center, so it is not counted. Agents can issue refunds through Stripe, but taking a customer payment in the conversation is not documented.

Google Cloud service for building chat and voice agents that combine deterministic flows with Gemini based generative agents, sold within the Gemini Enterprise for Customer Experience portfolio.

Vendor: Google (Google Cloud)

Models named: Google (Gemini)

Pricing: Pay per use: chat agents billed per request ($0.007 Flows, $0.012 Playbooks), voice agents billed per second of audio ($0.001 Flows, $0.002 Playbooks); data store storage billed above a 10 GiB free quota.

Notes: Google's compliance pages list the product under several names (Dialogflow, CX Agent Studio, Customer Experience Agent Studio), and no Google ISO/IEC 42001 scope found names any of them, so ISO 42001 is left unknown. No Google source was found on running the agent service on premises, in a customer cloud or air gapped, nor on outbound campaigns run by the agent. Payments rest on the prebuilt Make a payment component, which collects card details in the conversation and submits the payment through a customer configured webhook.

Conversational AI platform for building customer experience AI agents across messaging, web and voice channels.

Vendor: Jio Haptik Technologies Limited

Models named: OpenAI, Meta, Anthropic

Pricing: Not publicly documented

Notes: The trust center (trust.haptik.ai) shows an ISO/IEC 42001:2023 badge next to frameworks such as NIST AI RMF but no certificate was found, so iso42001 stays unknown, and no SOC 2 report is listed. The ISO 27001:2022 certificate (Alcumus ISOQAR, scope covering the Contakt and Interakt conversational AI products) runs to 9 November 2026, and the trust center lists HIPAA with a HIPAA report. The Arabic source is a help article last updated November 2022, and the homepage names GPT, Llama and Claude as models customers can experiment with, without stating a bring your own LLM option.

Platform for building, orchestrating and governing AI agents and conversational assistants, offered as SaaS on IBM Cloud or AWS or as software installed on premises.

Vendor: IBM

Models named: IBM watsonx.ai, OpenAI, Anthropic, Google, Azure OpenAI, Azure AI, AWS Bedrock, Mistral, Groq, xAI, OpenRouter, Ollama, Red Hat OpenShift AI

Pricing: Monthly plans: Essentials from $530 and Standard from $6,360 per month with caps on monthly active users and messages; Premium is custom priced.

Notes: IBM's former watsonx Assistant security page (which claimed ISO 27001, SOC 2 and PCI DSS) now redirects and no current IBM page naming Orchestrate in those scopes was found, while IBM's ISO 42001 announcement covers the Granite models, not Orchestrate. HIPAA is a HIPAA ready option on the Premium plan only, private cloud rests on a 2024 IBM Community post saying the Cloud Pak for Data edition runs on premises or on third party clouds, and the Arabic source is an IBM product manager's answer in the IBM Community. Live agent transfer works through a watsonx Assistant collaborator agent and is not supported in on premises deployments.

Platform for building, deploying and governing AI agents for customer service, employee support and process automation.

Vendor: Kore.ai

Models named: OpenAI, Anthropic, Azure OpenAI, Microsoft Foundry, Google, Vertex AI, AWS Bedrock, Cohere, Mistral AI, Groq, Fireworks, Together AI, Perplexity, DeepSeek, xAI, Ultravox

Pricing: Usage based: XO Standard plan published at $0.20 per conversation, Enterprise plan billed per 15 minute session under a custom contract; Agent Platform plans are Free, Team (per seat), Business (per seat plus usage) and Enterprise (custom).

Notes: The Artemis press release (May 2026) states Kore.ai is SOC 2 Type II, ISO 27001 and PCI DSS certified, and trust.kore.ai lists the same three under Audits and Certifications; HIPAA is not on the trust center, and Kore.ai calls itself "HIPAA aligned" in the release but states HIPAA compliance is built in on its healthcare page. No native payment capture inside the conversation is documented by Kore.ai itself. The official HQ remains Orlando, with a Bay Area strategic headquarters (San Mateo) opened in May 2026.

Blits.ai vs Kore.ai, in detail →

Conversational Cloud platform for messaging based customer engagement that combines AI agents and human agents.

Vendor: LivePerson (acquired by SoundHound AI, completed September 2026)

Models named: Google, OpenAI, Microsoft, Amazon, Anthropic, Cohere, IBM

Pricing: Bronze, Silver and Gold tiers priced by quote; proactive SMS and WhatsApp billed at provider rates plus a 15% handling fee

Notes: The LivePerson trust center lists ISO 27001:2022, SOC 2 Type 2, PCI and HIPAA and holds an ISO 27001 certificate, a SOC 2 Type 2 plus HITRUST report and a PCI DSS Attestation of Compliance, although the marketing security page still describes the ISMS as "aligned with" ISO 27001. CX Today reports LivePerson was removed from the 2026 Magic Quadrant, and generative AI features are moving from Azure OpenAI to Google Gemini during 2026 (Azure kept where needed for data residency). Arabic is listed for web messaging only, not the mobile SDK, and guardrails evidence is limited to LLM Gateway hallucination detection and PII masking.

Low code agent builder within Microsoft Power Platform for creating chat and voice agents and publishing them to Microsoft 365, web, mobile and messaging channels.

Vendor: Microsoft

Models named: OpenAI (GPT), Anthropic (Claude), xAI (Grok), Mistral, Azure AI Foundry models (bring your own)

Pricing: Consumption in Copilot Credits: prepaid packs of 25,000 credits for $200 per month, pay as you go via an Azure subscription, or prepurchase commit units.

Notes: PCI DSS is unknown because Microsoft's PCI DSS offering page lists Power Apps, Power Automate and Power BI but not Copilot Studio, and the Copilot Studio compliance page only says it is "compliant with or covered by" PCI DSS without naming an attestation. Anthropic, xAI and Mistral models are external models an admin must enable, and bring your own model via Azure AI Foundry applies to the prompt tool. Live agent handover needs Dynamics 365 Customer Service and outbound voice or SMS needs Customer Insights Journeys plus Contact Center; Copilot Studio is not among the 14 vendors in the 2026 Gartner Magic Quadrant as reported by CX Today.

AI agent platform for enterprise customer service that handles chat, email and voice interactions.

Vendor: Netomi, Inc.

Models named: OpenAI, Anthropic, Google

Pricing: Production pricing is customized per enterprise via private offer; AWS Marketplace lists a one month proof of concept at USD 100,000

Notes: Netomi's security page shows SOC 2 Type II, ISO 27001, HIPAA and PCI DSS only as badges without a stated certification status and shares attestations only through a Netomi representative, so ISO 27001, SOC 2 and PCI DSS are unknown, while HIPAA reflects the AWS Marketplace listing stating HIPAA compliance. The public website now carries little product detail, so scope evidence comes from the AWS Marketplace listing, the App Store description, a Gladly integration guide and press coverage. Privately held; raised a USD 110 million Series C led by Accenture Ventures in April 2026, and its website states new engagements are by referral.

Conversational AI platform for contact center voice and digital self service, including voice biometrics and fraud prevention.

Vendor: Omilia Natural Language Solutions

Pricing: Not publicly documented

Notes: The Omilia trust center holds an ISO/IEC 27001:2022 certificate, a 2025 SOC 2 Type 2 report (Schellman, December 2024 to November 2025) and a PCI DSS v4.0.1 Attestation of Compliance (assessed April to August 2026 on a specific OCP deployment), and its launch post names HIPAA among its certifications. The platform page offers multi tenant, exclusive tenant SaaS or Private Cloud models with tools for customers to build and manage their own deployment, and says the platform uses trusted external and in house LLMs without naming providers. Corporate HQ is Larnaca, Cyprus with a delivery center in Athens, and the published language graphic does not show Arabic but ends with "and more".

AI agent management platform for building, testing and running customer service agents, primarily for contact center voice.

Vendor: Parloa

Models named: OpenAI, Google Gemini

Pricing: Consumption based: per minute for voice and per interaction for chat, volume packages, custom quote

Notes: Founded in Berlin; press materials mention a US HQ in Manhattan, so HQ country could also be read as United States. Parloa states it is ISO 27001:2022 certified with SOC 2 Type II attestation and PCI DSS certification, and its trust center (trust.parloa.com) lists the matching certificate, reports and PCI DSS AOC, although some Parloa pages describe PCI DSS and HIPAA as alignment and no BAA offer was found. Parloa states 130+ languages without naming Arabic, runs on Microsoft Azure in vendor operated regional zones, and delivers outbound together with partner Alvaria.

Voice AI agents that answer customer service phone calls for enterprises, extended to chat and messaging channels.

Vendor: PolyAI

Models named: PolyAI (Raven), OpenAI, Amazon Bedrock (Anthropic Claude, Amazon Nova)

Pricing: Not publicly documented

Notes: PCI DSS is unknown because PolyAI's compliance docs only state a commitment to comply with PCI DSS (a healthcare marketing page claims full certification without detail), and card payments during calls run through a PCI Pal integration whose Level 1 certification belongs to PCI Pal. HIPAA rests on vendor statements (HIPAA "out of the box" on the healthcare page, systems designed to meet HIPAA requirements in the docs), with no public mention of a BAA. The default model is the proprietary Raven 3.5, with OpenAI and Amazon Bedrock models and customer LLM endpoints also selectable; Arabic is a Raven response language, but ASR support is described as provider dependent.

Blits.ai vs PolyAI, in detail →

Developer framework and platform for building chat and voice AI agents that run in the customer's own infrastructure.

Vendor: Rasa

Models named: OpenAI, Azure OpenAI, Amazon Bedrock, Mistral, Google Gemini (AI Studio), Hugging Face Inference Endpoints, Anthropic, Cohere, Together AI, Groq, Self hosted (vLLM, Ollama, llama.cpp)

Pricing: Free Developer Edition (up to 1,000 external conversations per month); Enterprise licensed annually by conversation volume, quote on request

Notes: Rasa's security page states it is not a SaaS application, but its pricing page and docs also offer a Managed Service in which Rasa hosts the deployment. No certification is claimed: the security page says controls are aligned with ISO 27002 and a comparison page mentions "SOC 2 Type II compliance" without an attestation, so all certifications stay unknown. Air gapped deployment is stated on Rasa product pages but not described in the docs, human handover is a pattern the customer implements rather than a ready connector, and Rasa names San Francisco as HQ with engineering in Berlin.

Developer platform for building, testing and deploying AI voice agents for phone calls, with text chat and SMS agents.

Vendor: Retell AI, Inc.

Models named: OpenAI, Anthropic, Google

Pricing: Pay as you go: voice agents per minute ($0.07 to $0.31/min, built from LLM, voice infrastructure, TTS and telephony rates), chat agents per message; custom enterprise pricing.

Notes: Private cloud (VPC), on premises and air gapped options come from Retell marketing pages (October 2026); an older Retell community support thread said on premises was not offered and hosting was single region US, so these are vendor claims without public technical documentation. The trust center holds SOC 2 Type 2 reports, an ISO 27001:2022 certificate and a BAA, but no PCI DSS document, and no in call payment feature was found. Arabic is documented with a single locale (ar-SA), and HQ city varies by source (San Carlos, Redwood City, San Francisco).

AI agent platform built into Salesforce CRM for service, sales and employee use cases across messaging, web, mobile and voice channels.

Vendor: Salesforce

Models named: OpenAI, Anthropic, Google (Gemini)

Pricing: Flex Credits at $500 per 100,000 credits (20 credits per standard action, 30 per voice action), or $2 per conversation, or per user add ons from $125 per user per month.

Notes: Salesforce lists a PCI Attestation of Compliance covering Data Cloud, Einstein Platform and Agentforce, while its Einstein Platform security document says card numbers may not be submitted to the covered services. Arabic is announced for Service Agent and Employee Agent, whereas Agentforce Voice is listed as available in English only. Payments stay unknown: in chat payment is announced for the B2B Carter for Buyers agent (GA January 2027) and an Instant Checkout integration in ChatGPT was announced in 2025, but no shipped in conversation payment flow was confirmed.

AI agents for customer experience that handle conversations across voice, chat, email and messaging channels.

Vendor: Sierra Technologies, Inc.

Pricing: Outcome based: customers pay when the agent completes an agreed outcome (for example a resolved conversation); no rates published on sierra.ai.

Notes: Sierra says it combines 15+ frontier, open weight and proprietary models and switches between LLM providers, but does not name providers publicly, and its public pages do not describe the hosting model (a Sierra job posting mentions deployments in customer owned cloud environments, not treated as sufficient evidence). Sierra states 59 languages across voice, chat, email and WhatsApp but publishes no language list, so Arabic stays unknown; it is absent from the 14 vendor 2026 Gartner MQ as listed by CX Today.

Blits.ai vs Sierra, in detail →

Voice focused AI agent platform (Amelia 7 and OASYS) for customer and employee service across phone, chat, in vehicle and device channels.

Vendor: SoundHound AI, Inc.

Models named: OpenAI, Azure OpenAI, Google Gemini, Anthropic, Mistral, xAI Grok, Amazon Bedrock, SoundHound Tempus (proprietary)

Pricing: Not published; contact sales

Notes: The SoundHound trust center shows SOC 2, ISO 27001, HIPAA and PCI DSS logo badges with only the text "SOC 2 / ISO" and no stated audit type, scope or status, so SOC 2, ISO 27001 and PCI DSS are unknown; HIPAA rests on the healthcare page calling its agents HIPAA compliant, with no BAA mentioned. On premises is unknown because the Amelia 7.1 release note covers custom on premises LLMs rather than the platform itself, and current OASYS pages describe hosting in SoundHound's cloud. Outbound and payments appear only as use case examples (reminders, drive thru and in car ordering), not as documented features, so both stay unknown.

Customer experience management suite whose Sprinklr Service product includes AI agents for voice, chat, email and social channels.

Vendor: Sprinklr, Inc.

Models named: Azure OpenAI, OpenAI, Google Vertex AI, Amazon Bedrock, Anthropic, xAI, Sprinklr in house LLM

Pricing: Contract based on term and committed quantity, with usage based charges for overage (AWS Marketplace listing for the Unified CXM platform)

Notes: The trust page states Sprinklr maintains SOC 1 Type II, SOC 2 Type II, PCI DSS and ISO 27001 certifications and FedRAMP authorization, hosted on AWS, Microsoft and Google infrastructure in the United States and Europe; HIPAA is not listed there. Bring your own key covers OpenAI, Azure OpenAI, Google Vertex, Amazon Bedrock, Anthropic and xAI, while Sprinklr managed keys cover Azure OpenAI, Google Vertex, Amazon Bedrock and an in house LLM; Sprinklr is publicly listed (NYSE: CXM). The payments value rests on voicebot documentation listing making payments as a telecom use case, without detail on the payment flow. Payments is not counted: the source mentions payments as a use case without describing a payment flow inside the conversation.

Agentic AI platform that automates contact center voice and digital conversations on top of existing contact center systems.

Vendor: Teneo.ai (Teneo AI AB, formerly Artificial Solutions)

Models named: OpenAI, Anthropic, Azure OpenAI, AWS Bedrock

Pricing: Not publicly documented

Notes: The security center states ISO 27001 certification for the whole organization and SOC 2 Type I and II reports (Security principle) from 2024 onwards, but only says Teneo can be implemented in HIPAA compliant use cases, with no compliance statement or BAA found, so hipaa is unknown. Teneo 8.0 documentation has a section of features "only available for customers on premise", which supports on_premises, while current marketing pages describe SaaS only. Teneo claims 86+ languages but the documented language list does not name Arabic, so arabic stays unknown rather than no.

Enterprise AI platform (Business AI Cloud) for building AI agents and applications, including customer self service across voice and chat.

Vendor: Uniphore Technologies Inc.

Models named: OpenAI (GPT), Anthropic (Claude), Google (Gemini), Mistral, Meta (LLaMA)

Pricing: Not publicly documented

Notes: Arabic ASR was reported in October 2022 coverage of the X platform (kmworld.com) but no current Uniphore source naming Arabic was found; the SafeBase trust center (trust.uniphore.com) was unavailable, so certifications rest on the Uniphore security page (ISO 27001:2022 and SOC 2 Type II certified, Level 1 PCI DSS service provider with AOC). Private cloud is based on SiliconANGLE's 'any cloud or on premises infrastructure' wording plus Uniphore's own 'On Premises and Multi Cloud' positioning, and HIPAA is stated as adherence to HIPAA standards rather than a certification. Human handover, outbound and payments were not documented on current Uniphore pages.

Hosted platform for designing, building and deploying chat and phone AI agents, mainly for customer support.

Vendor: Voiceflow

Models named: OpenAI, Anthropic, Google, AWS Bedrock

Pricing: Usage based credits (LLM calls, messages, voice minutes); business pricing on request

Notes: Voiceflow states it is ISO/IEC 27001:2022 certified and SOC 2 Type II with the audit report in its trust center; HIPAA means it supports HIPAA compliant deployments, with BAA terms handled case by case. No Voiceflow page states private cloud or on premises hosting and a competitor calls it cloud only, so self hosting stays unknown. Guardrails are rules set in the agent's global prompt, and payments appear only as a banking use case on a marketing page, so payments stay unknown.

Enterprise AI platform with locally embedded deployment teams that builds and runs AI agents for customer service and internal processes in regulated industries, with a focus on non English speaking markets.

Vendor: Wonderful

Models named: Google (Gemini), Anthropic (Claude)

Pricing: Not publicly documented

Notes: Founded in Tel Aviv in 2025, but its Series C release (September 2026) is datelined Amsterdam and the operating entity is Wonderful AI B.V.; Salesforce joined that round as a minority investor, not an acquirer. Its trust center lists an ISO 27001/17/18 certificate, an ISO 42001 AIMS certificate, a PCI DSS v4.0.1 AOC and a 2026 SOC 2 Type 2 report (documents gated), and the vendor describes its on premises option as completely air gapped. No model providers are named on its own site, so they come from a Google Cloud case study and the Anthropic Agent Builder launch, and no live handover, in conversation payment or visual builder is documented.

AI agent platform for automating customer service across voice, chat, email and messaging channels.

Vendor: Yellow.ai

Models named: OpenAI, Azure OpenAI, AWS Bedrock, Anthropic

Pricing: Per resolution: Free plan includes 500 resolutions per month, then $0.99 per resolution; Enterprise is custom; AWS Marketplace lists 12 month contracts at $10,000 (Basic) and $25,000 (Standard).

Notes: Deployment beyond SaaS rests only on Yellow.ai's own AWS Marketplace listing, which names public cloud, private and on premises options without saying the platform runs in the customer's cloud account (so private cloud is unknown), and one third party review claims no self hosted option. The trust center states ISO 27001 certification, a SOC 2 Type II attestation, HIPAA compliance as a Business Associate and a PCI DSS service provider AOC scoped to the North America region. The about page lists San Mateo, California as the first office without labelling it headquarters (the company was founded in Bengaluru), and Yellow.ai dropped from Challenger (2025) to Niche Player (2026) in the Gartner MQ.

Blits.ai vs Yellow.ai, in detail →