Customer service agents, in the vendors’ own words
Every cell quotes the vendor. Checked 18 Nov 2026: 61 of 61 quotes still on their pages. How
Customer service agents, feature by feature
| Test before launch | ||||||
|---|---|---|---|---|---|---|
| Follows your procedures | ||||||
| Acts in your systems | ||||||
| Chat, email and messaging | ||||||
| Answers phone calls | ||||||
| Works with other help desks | ||||||
| Audit logs | ||||||
| Data residency | ||||||
| Not used for training | ||||||
| SOC 2 and ISO 27001 | ||||||
| HIPAA BAA |
- Yes
- Partly
- No
- Not documented
- Words in cells are each vendor’s own name for the feature. Click a cell for its source.
Can builders run the agent against test cases or simulated conversations before it goes live? Simulations run saved test cases as multi-turn conversations, with pass or fail results before changes go live.
Run a full simulation suite before deploying changes, generating clear pass/fail metrics to share with stakeholders and support go/no-go decisions.
Can builders run the agent against test cases or simulated conversations before it goes live? Simulations test agent behavior across channels before release and after every update, and can run on a schedule.
Simulations, Decagon's integrated testing suite, help teams validate agent behavior across channels before deploying to production and with every subsequent update.
Can builders run the agent against test cases or simulated conversations before it goes live? Simulations run full test conversations against a Procedure; Evals test the whole Fin setup.
Simulations allow you to validate Fin Procedures, build confidence in your automation, and catch problems before they impact your customers.
Can builders run the agent against test cases or simulated conversations before it goes live? The Test tab runs simulated customer queries to check answers, workflow triggers and fallbacks before launch.
The Test tab in AI Agent Studio is a vital tool for simulating real customer conversations.
Can builders run the agent against test cases or simulated conversations before it goes live? Simulations run test conversations scored against set outcomes; a test suite catches regressions before release.
Simulate conversations and evaluate your agent’s performance against defined outcomes.
Can builders run the agent against test cases or simulated conversations before it goes live? A test widget runs full or single-dialogue test chats; it cannot test CRM actions or escalations.
After you create or update an AI agent, you can test that it functions as intended before making it live.
Can staff write step-by-step procedures in plain language for the agent to follow? Playbooks are step-by-step, but built from typed steps (ask, set, run, branch) plus written guidelines, not free text.
Playbooks are structured, step-based workflows that guide AI Agents through complex, multi-step conversations.
Can staff write step-by-step procedures in plain language for the agent to follow? CX teams write AOPs in natural language, like staff SOPs; engineers can add code and Git versioning.
Instead of complex SDKs or costly professional services, you can define agent behavior in natural language, the same way you train human agents with SOPs.
Can staff write step-by-step procedures in plain language for the agent to follow? Staff write Procedures in natural language, with optional conditions and code for strict steps.
Write your processes in natural language, just like you would when training a teammate, or let AI draft them for you.
Can staff write step-by-step procedures in plain language for the agent to follow? Writing workflows in natural language is in early access; the generally available builder is a visual canvas.
Instead of constructing workflows block by block, administrators can describe business processes using natural language while continuing to use structured workflow capabilities such as triggers, API actions, collected inputs, property references, group assignments, and agent handoffs.
Can staff write step-by-step procedures in plain language for the agent to follow? Teams write step-by-step Journeys without code, or have AI draft them from existing operating procedures.
Define step-by-step workflows for your agent from scratch, or use AI to instantly generate journeys from your existing operating procedures.
Can staff write step-by-step procedures in plain language for the agent to follow? Admins describe how to resolve an issue in step-by-step instructions that the AI agent follows.
Create generative procedures to guide an AI agent through customer conversations with flexible, step-by-step instructions.
Can the agent do things like issue a refund or change an order by calling your other systems? API, code and MCP tools let the agent call your systems, for example to check eligibility and complete a refund.
Tools let an AI Agent do things beyond the conversation — call an external system, run logic, and return exactly what it needs to answer or act.
Can the agent do things like issue a refund or change an order by calling your other systems? Agents issue refunds or store credit, process returns and update inventory in connected systems.
Process returns and exchanges against policy rules, issue refunds or store credit, and update inventory automatically.
Can the agent do things like issue a refund or change an order by calling your other systems? Data connectors call your APIs so Fin can process refunds or change accounts and subscriptions.
Execute real-time actions like processing refunds, updating accounts, or modifying subscriptions in external systems using Data connectors.
Can the agent do things like issue a refund or change an order by calling your other systems? Workflows call APIs and update other systems, for example to cancel orders or change subscriptions.
Workflows extend your AI Agent’s ability to think, reason, and act—automating complex workflows like order cancellations or subscription updates.
Can the agent do things like issue a refund or change an order by calling your other systems? The agent connects to order systems and CRMs to act for the customer, such as exchanges or rebookings.
Securely connect to your systems of record—from order management systems to CRM—so your agent can take action on your customer’s behalf, like processing an exchange or changing a reservation.
Can the agent do things like issue a refund or change an order by calling your other systems? Custom actions call outside systems through APIs to read or update data; Growth plan and above.
Create custom actions that let auto assist, AI agents, and action flows call external systems to retrieve or update information.
Can the same agent answer on web chat, email and messaging apps such as WhatsApp? One AI Agent runs on web and app chat, email, voice and social apps such as WhatsApp.
You’ll learn how to set up and manage Chat , Voice , Email , Social , and Third-party channels.
Can the same agent answer on web chat, email and messaging apps such as WhatsApp? The same agent runs on web chat, SMS and WhatsApp, with separate email and voice products.
Handle inbound requests end-to-end across web live chat, SMS, WhatsApp, and more.
Can the same agent answer on web chat, email and messaging apps such as WhatsApp? Fin deploys on Messenger chat, email, phone, WhatsApp, SMS, Facebook Messenger, Instagram and Slack.
You can deploy Fin AI Agent to all your inbound support channels.
Can the same agent answer on web chat, email and messaging apps such as WhatsApp? Chat AI Agents cover web chat, WhatsApp and social apps; a separate Email AI Agent answers email.
Chat AI Agents handle real-time conversations across messaging channels such as, WhatsApp, Web Chat, Facebook, Instagram, and more.
Can the same agent answer on web chat, email and messaging apps such as WhatsApp? One agent serves web and app chat, WhatsApp, Apple Business Chat, SMS, email and voice.
Power chat conversations across your website, mobile app, WhatsApp, Apple Business Chat, and SMS.
Can the same agent answer on web chat, email and messaging apps such as WhatsApp? AI agents answer on Zendesk messaging (web, WhatsApp and other social channels), email and voice.
AI agents with agentic AI are available on messaging, email, and voice channels.
Can the agent talk with customers on the phone? Voice agents answer calls and can route to your existing contact center phone system.
With Ada Voice, your customers can engage with your AI Agent over the phone, helping them to resolve their issues faster.
Can the agent talk with customers on the phone? Voice agents take inbound calls, make outbound calls and transfer to humans with a summary.
Voice AI agents built for natural, multilingual conversations and customized to your brand.
Can the agent talk with customers on the phone? Fin Voice answers phone calls but is limited to select customers working with sales, on custom pricing.
Fin Voice is currently limited to select customers partnered with our Sales team, and is supported on workspaces hosted in the US, EU, and Australia.
Can the agent talk with customers on the phone? Phone agents come from third-party Marketplace apps (Synthflow, SquawkVoice), not Freddy AI Agent itself.
Voice AI Agents are now available via Freshcaller Marketplace apps—Synthflow and SquawkVoice.
Can the agent talk with customers on the phone? Voice agents handle inbound and outbound calls and can replace IVR menus.
Replace rigid IVR menus with an empathetic voice agent that understands customer context, leverages memory for personalized experiences, and delivers accurate, real-time resolutions while representing your brand.
Can the agent talk with customers on the phone? Voice AI agents answer calls and hand off to a human; they need Zendesk voice (Talk).
Voice AI agents enable intelligent, conversational phone support by automating end-to-end customer voice interactions.
Can it run on top of help desks from other vendors, such as Zendesk or Salesforce? Ada hands off into Zendesk, Salesforce, ServiceNow, Genesys, Gorgias, Freshchat and other platforms.
Your AI Agent can integrate with external support systems like Zendesk and Salesforce to seamlessly transfer conversations to human agents.
Can it run on top of help desks from other vendors, such as Zendesk or Salesforce? Decagon syncs tickets with Salesforce, Zendesk and Intercom and hands off inside them.
Sync support tickets and tap into customer data from applications like Salesforce, Intercom, Zendesk, and more.
Can it run on top of help desks from other vendors, such as Zendesk or Salesforce? Fin runs on Zendesk tickets and messaging, Salesforce Service Cloud, Freshdesk and HubSpot.
Fin connects directly to your Zendesk tickets and messaging, so you can deploy the highest performing AI Agent and keep your team on Zendesk.
Can it run on top of help desks from other vendors, such as Zendesk or Salesforce? Searched the Freshdesk and Freddy AI help centers and Freshworks product pages; Freddy AI Agent is documented only inside Freshworks.
No statement from Freshworks on its own pages. Searched the Freshdesk and Freddy AI help centers and Freshworks product pages; Freddy AI Agent is documented only inside Freshworks.
Can it run on top of help desks from other vendors, such as Zendesk or Salesforce? Sierra connects to existing contact centers and hands off to them; public pages name no specific help desk.
Agents built on Sierra already connect to knowledge sources, systems of record, and contact centers to operate with accuracy, take action, and hand off to support teams when needed.
Can it run on top of help desks from other vendors, such as Zendesk or Salesforce? Zendesk's own AI agents run on Zendesk; the separately sold Forethought AI Agents work on other platforms.
Forethought AI Agents by Zendesk work seamlessly within Zendesk and across any platform, making it easy for customers to deploy best-in-class AI agents within their existing technology stacks.
Can admins export or query a log of who did what? The audit log records who changed the AI Agent's setup, when and how; it can also be read by API.
The Audit log records configuration changes made to your AI Agent, giving administrators a chronological, searchable history of who changed what, when, and through which interface.
Can admins export or query a log of who did what? Teams see audit logs through Agent Workbench for debugging; no admin export of user activity is described.
That’s why we built Agent Workbench, an AI chat assistant that accelerates debugging with consolidated access to audit logs, agent reasoning, tool errors, and more.
Can admins export or query a log of who did what? Activity logs record who changed settings or exported data; available on all plans, via API and export.
Teammate activity logs let you monitor important actions taken by teammates in your workspace, such as updating customer profiles, managing settings, exporting data or changing inbox configurations.
Can admins export or query a log of who did what? The Audit Log shows admins who changed which helpdesk setting and when.
Audit Log helps Admins oversee these changes made in the account.
Can admins export or query a log of who did what? Agent changes are versioned with full history; no admin audit log export is described publicly.
Journeys, configuration, and simulations are versioned automatically, so teams can see exactly what changed and when.
Can admins export or query a log of who did what? The account audit log records who changed what and exports to CSV, but only on Enterprise plans and above.
The audit log allows customers on Enterprise plans and above to view various changes in their Zendesk account since the account was created.
Can the company choose the region where its data is stored? Ada docs show US, EU and Canada agent instances but never say customers choose; ada.cx security pages block automated checks.
No statement from Ada on its own pages. Ada docs show US, EU and Canada agent instances but never say customers choose; ada.cx security pages block automated checks.
Can the company choose the region where its data is stored? Decagon says data residency is built in but does not list the regions you can choose.
Layered guardrails monitor 100% of conversations in real time, with configurable PII redaction and data residency built in.
Can the company choose the region where its data is stored? Workspaces are hosted in the US, the EU or Australia.
Data is hosted in the US, EU, or Australia based on your residency needs, with redundant systems architected for continuous uptime across regions.
Can the company choose the region where its data is stored? You pick a data center at sign-up, but some Freddy AI processing for India, Australia and Middle East accounts runs in the US.
You can choose your preferred data location when you sign up for your account.
Can the company choose the region where its data is stored? Searched sierra.ai product, trust, industry and blog pages; no statement on choosing a data region. Docs and trust center need sign-in.
No statement from Sierra on its own pages. Searched sierra.ai product, trust, industry and blog pages; no statement on choosing a data region. Docs and trust center need sign-in.
Can the company choose the region where its data is stored? Region choice needs Suite Professional or above; AI agent accounts made before March 10, 2026 get only EEA or US.
Only AI agents accounts created on or after March 10, 2026 support data Hosting locality.
Is business data excluded from model training by default? Searched Ada docs: LLM providers keep no data (zero retention), but nothing says whether data trains models.
No statement from Ada on its own pages. Searched Ada docs: LLM providers keep no data (zero retention), but nothing says whether data trains models.
Is business data excluded from model training by default? Model providers keep and train on nothing; Decagon doesn't say whether its own post-trained models use your data.
Decagon enforces zero-data retention with all AI providers like OpenAI and Anthropic, ensuring no conversation data is stored or used for training.
Is business data excluded from model training by default? Third-party AI providers may not train on your data, but Intercom fine-tunes on anonymized data unless you opt out.
Fin can use anonymized customer data for model fine-tuning, and you can opt out at any time, with data deleted within 30 days.
Is business data excluded from model training by default? Third-party models don't train on your data, but Freshworks trains its own models on AI users' data unless they opt out.
Freshworks only uses data from Customers who are using AI features for AI Training.
Is business data excluded from model training by default? Sierra says customer data is never used to train models.
Your data is never used to train models, and we use industry-standard best practices to ensure its security.
Is business data excluded from model training by default? Third-party LLMs never train on your data; Zendesk's own models may train on sanitized data, which customers control.
Zendesk uses pre-trained models, and no third-party will use your inputs to train their models or otherwise improve their services.
Does the vendor state SOC 2 Type II and ISO 27001 for this product? Ada docs have no SOC 2 or ISO 27001 statement; ada.cx trust pages block automated checks, so they can't be cited.
No statement from Ada on its own pages. Ada docs have no SOC 2 or ISO 27001 statement; ada.cx trust pages block automated checks, so they can't be cited.
Does the vendor state SOC 2 Type II and ISO 27001 for this product? Decagon lists SOC 2 Type II and ISO 27001, plus HIPAA, GDPR and PCI.
Compliance certifications SOC 2 Type II, HIPAA, GDPR, PCI, and ISO 27001
Does the vendor state SOC 2 Type II and ISO 27001 for this product? Intercom states SOC 2 Type II and ISO 27001 for Intercom and Fin, plus ISO 42001 and AIUC-1.
Intercom and Fin hold SOC 2 Type II and HIPAA compliance along with ISO 27001, ISO 27018, ISO 27701, and ISO 42001 certification
Does the vendor state SOC 2 Type II and ISO 27001 for this product? Freshdesk is audited each year for ISO 27001, ISO 27701 and SOC 2 Type 2.
Freshdesk is audited annually by independent audit firms for ISO 27001, ISO 27701, SOC 2 Type 2, and VAPT.
Does the vendor state SOC 2 Type II and ISO 27001 for this product? Sierra holds a SOC 2 Type II attestation and ISO 27001 and ISO 42001 certifications, plus AIUC-1.
This certification complements Sierra’s existing SOC 2 Type II attestation, and ISO 27001 and ISO 42001 certifications.
Does the vendor state SOC 2 Type II and ISO 27001 for this product? Zendesk holds a SOC 2 Type II report and ISO 27001:2022 certification; the page does not list products in scope.
We undergo routine audits to receive updated SOC 2 Type II reports, available upon request and under NDA.
Will the vendor sign a HIPAA business associate agreement that covers this product? Docs treat HIPAA eligibility by channel and exclude email; no readable page states BAA terms.
Email is currently not HIPAA eligible.
Will the vendor sign a HIPAA business associate agreement that covers this product? Decagon states HIPAA compliance, but no public page says it signs a BAA.
SOC 2 Type 2, HIPAA, GDPR, and CCPA compliance, configurable PII redaction, and layered guardrails that run before, during, and after every conversation.
Will the vendor sign a HIPAA business associate agreement that covers this product? Intercom signs a BAA with eligible customers; it covers Fin AI Agent and Fin Voice.
Fin becomes a Business Associate of eligible customers via a BAA.
Will the vendor sign a HIPAA business associate agreement that covers this product? Freshworks says Freddy AI products are not part of its HIPAA configuration guides.
Freddy AI products are not currently part of any Freshworks HIPAA Configuration Guides.
Will the vendor sign a HIPAA business associate agreement that covers this product? Sierra lists HIPAA among its compliance standards, but no public page says it signs a BAA.
Sierra is committed to maintaining the highest compliance standards for our customers, including SOC 2, HIPAA, GDPR, PCI, FedRAMP High, CCPA, CSA STAR, ISO 27001, and ISO 42001.
Will the vendor sign a HIPAA business associate agreement that covers this product? BAA through Advanced Compliance on Suite Professional or above; AI Agents - Advanced is a listed HIPAA-enabled add-on.
Advanced Compliance lets you enter a Business Associate Agreement (BAA) for HIPAA or an HDS Terms Exhibit for French health data hosting
Latest change, 18 May 2026: Zendesk AI agents. Zendesk splits AI resolutions into tiers. All changes
Who sells what
A few companies sell in most of these categories, so start from the row you already pay for. Everyone else sells one kind.