Telnyx vs Vonage (2026)

Assistants call outside systems with webhook tools, which can run asynchronously with filler messages while the request completes, attach MCP servers for whole tool catalogs, and connect enterprise integrations for Salesforce, Zendesk and Jira. Call control tools transfer the caller, hand off with SIP Refer, send DTMF tones to another phone system and hang up, client side tools run actions in the web widget, and keys for outside providers are kept as write only integration secrets. Tools live in a shared library that several assistants and workflow nodes can reuse.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant.mdread 2026-10-11

A Webhook node calls any outside API mid conversation with any of the five standard HTTP methods, carries authorization in custom headers, sends JSON and XML bodies as well as HTML, plain text or form content built from conversation parameters, and maps fields from the response into parameters the flow uses next.

Success and failure paths branch on status codes or code classes, requests can run asynchronously, and timeouts run from 10 to 25 seconds. Salesforce action nodes get records through SOQL and create or update records on any object the team names, after a connection set up from the API dashboard. An email connector and custom code nodes round out the set, so an agent can update a booking, open a case or check an order while the customer waits.

Sourcestudio.docs.ai.vonage.com/voice/nodes/integrations/webhook.mdread 2026-10-11

Conversation workflows store a directed graph on the assistant. Prompt nodes run the model with their own instructions and optional model and voice overrides, speak nodes play a fixed message, and tool nodes run one shared tool and advance.

Edges move between nodes on a natural language condition the model judges, a deterministic comparison against variables, or a default, and a tool node's HTTP status code can steer the next edge.

An edge can also hand the conversation to another assistant, so a triage assistant can route billing, technical and sales callers to specialists that share the conversation's context.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/workflows.mdread 2026-10-11

Agents run as flows of nodes on a drag and drop canvas, where collect input, condition, webhook, Salesforce, Q&A, routing and set parameter nodes chain into multistep business logic, and inbound and outbound events start different parts of the same agent. The agentic NLU engine adds a hub and branch pattern.

An agentic classification node uses a language model to match the caller to an intent, an agentic capture node gathers that intent's details, and the agent picks its next move, such as asking a question or calling an API, sending the caller back to the hub when the intent changes. That engine is gated, English only and for voice agents, so most flows run on intent based routing designed in advance.

Sourcedeveloper.vonage.com/en/ai-studio/overviewread 2026-10-11

Calls reach an assistant on an assigned Telnyx number or over SIP, outbound calls start from the portal, the API or the TeXML ai_calls endpoint, and an embeddable widget carries voice and chat on a web page. SMS conversations run through the assistant SMS chat API, with MMS during a call when messaging is on. At the start of each conversation a dynamic variables webhook can load caller details and memory settings, and custom SIP headers or the outbound request can pass variables in.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant.mdread 2026-10-11

Four agent types cover telephony, WhatsApp, SMS and HTTP for web and app chat, and each agent listens for inbound or outbound events, with more events added later. Outbound voice, WhatsApp and SMS sessions start at one per second by default, with three or five a second on approval, and voice agents can call SIP endpoints. HTTP agents start through an init call and advance with a step call carrying the user's text and parameters, so any web or mobile front end can drive the same agent. Phone numbers bought on the Vonage dashboard link to an agent at publish time.

Sourcestudio.docs.ai.vonage.com/ai-studio/create-a-new-agent.mdread 2026-10-11

Each assistant can use named knowledge bases built from uploaded files or a URL, kept with the account and attached in the assistant's configuration, so one knowledge base can serve several assistants. Retrieval during a call is part of the voice engine's orchestration and included in the per minute price, while the price estimator carries a separate agent knowledge base add on. Telnyx does not say how the content is chunked or indexed.

Sourcetelnyx.com/pricing/conversational-airead 2026-10-11

Knowledge AI answers from the customer's own content through retrieval over Google Gemini models. Sources come in as PDF, text or HTML files, single web pages or public cloud storage links, are grouped into indexes shared by every agent under the same API key, and are queried through a Q&A node in the flow.

Replacing a source in an index updates live agents without republishing them, and an index tester shows which source produced an answer. Each master account includes a 100MB knowledge base, files run up to 10MB for PDF and 5MB for HTML or text, and tables and text inside images are not indexed.

Sourcestudio.docs.ai.vonage.com/ai-studio/knowledge-ai.mdread 2026-10-11

Memory brings earlier conversations into a new one. The dynamic variables webhook returns a query in the same form as the List Conversations endpoint, typically the caller's phone number, a time window or custom metadata, plus the IDs of insights to include, and the assistant starts with those conversations and insight results in context, across voice and messaging.

The webhook has one second to answer before the call carries on without memory. What comes back is stored conversation history picked by the customer's query. There is no separate memory store, and Telnyx sets out no retention period or way to delete one person's history.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/memoryread 2026-10-11

User parameters carry what an agent collects from one conversation to the next. Values are saved automatically against the caller's phone number, reused across sessions and across voice, WhatsApp and SMS, and shared by every agent under the same API key, so a collect input node can skip a question already answered. Only the last value for each parameter is kept, and HTTP agents are not covered. Vonage gives no retention period for these values and no way to erase one caller's stored history. Agentic nodes carry their own memory settings inside a single call.

Sourcestudio.docs.ai.vonage.com/ai-studio/users.mdread 2026-10-11

Transfer and SIP Refer tools pass a caller to a person or another phone system, with voicemail detection on transfer, and handoff tools move a conversation between assistants. Releases stay under the team's control. A new version is saved with notes, traffic rules send selected callers or a percentage to it, and rollback returns all traffic to the main version at once. No step lets a person approve a tool call or a reply before the assistant acts.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/version-testing-traffic-distribution.mdread 2026-10-11

Live agent routing hands WhatsApp, SMS and web chat conversations to people, with WebSocket connections for HTTP agents, and voice agents route calls onward. Agentic nodes ship with built in safety guardrails and detect frustration from repeated failures, escalating on their own, and a global intent such as talk to an agent can be set up. Edits stay in an editor version until someone publishes, and earlier versions can be restored. Nothing lets a person approve an action or a reply before the agent sends it.

Sourcestudio.docs.ai.vonage.com/ai-studio/agentic-nlu-ai-engine.mdread 2026-10-11

Business accounts sign in to the Mission Control Portal through SAML single sign on with Okta, Azure Active Directory, Google, OneLogin, Auth0 or LastPass. Telnyx holds ISO 27001:2022 and ISO 27701, a SOC 2 Type II report under NDA covering voice, messaging, video and wireless, and HIPAA coverage for voice, messaging, fax and AI inference with a BAA available, plus PCI DSS for its own billing. The SOC 2 scope does not name AI assistants, and Telnyx also claims a SOC 2 Type III, a report type the AICPA does not issue.

Sourcetelnyx.com/ai/compliance.mdread 2026-10-11

AI Studio holds ISO 27001 and HIPAA compliance, with SOC 2 Type II pending alongside PCI DSS compliance and a CSA listing, while Vonage's core SMS and voice APIs already hold ISO 27001 and a SOC 2 Type II report. Account owners invite team members to the API dashboard and assign each one permissions by checkbox, changing them later from a manage column, and sources and agents are scoped to API keys and sub accounts. Vonage's listings for AI Studio vary between its certification table and its product section, and they say nothing of single sign on for the dashboard.

Sourcedeveloper.vonage.com/en/getting-started/concepts/regulatory-certificationsread 2026-10-11

Every conversation is stored and listable through the conversations API, transcripts show which workflow node produced each assistant message, and the last tool call's HTTP status is kept as a variable. Insight groups run structured and unstructured analysis on each conversation and can post results to a webhook, with Telnyx managed insights alongside custom ones. Version notes record what changed in each release, and Telnyx does not mention an audit trail of who changed an assistant's configuration.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/workflows.mdread 2026-10-11

The call log shows every session with its transcript, the parameters collected underneath it and a flow path marking each node as success, warning or error, plus an overall status such as completed, no answer or failed. A tags report counts sessions by tag, and the Vonage Contact Center dashboard tracks resolved, escalated and abandoned sessions.

An audit log records each change to an agent, its type such as an edit, a publication or a deletion, the time and the user who made it. Transcripts are kept for 90 days and telephony recordings for 30, and an Insights API exposes transcripts and recordings to other systems.

Sourcestudio.docs.ai.vonage.com/ai-studio/reports.mdread 2026-10-11

Voice AI agents run in the United States, the primary data center, or in the European Union with EU data residency, where data does not leave the EU. AI inference is also offered in Asia Pacific with regional processing, though voice agents are not offered there, and Canada has neither. The assistant's language model can also point at the customer's own OpenAI compatible endpoint, such as one on AWS Bedrock, Azure OpenAI or a self run vLLM server, which keeps inference inside an environment the customer controls.

Sourcetelnyx.com/ai/compliance.mdread 2026-10-11

Each agent is created in either the US or Europe, and the choice carries through to its API host, with separate US and EU studio endpoints, so a team can keep European conversations on European infrastructure agent by agent. Knowledge AI indexes sit under the same API key as the agents that query them. AI Studio runs as a hosted service only, with no private cloud or on premises edition.

Sourcestudio.docs.ai.vonage.com/ai-studio/create-a-new-agent.mdread 2026-10-11

New assistants start from a template in the portal, with a blank template as the usual starting point, and Telnyx managed insights come ready to attach to any assistant. A shared tools library lets one set of tools serve many assistants. Telnyx offers no catalog of finished agents for support, sales or scheduling.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/no-code-voice-assistant.mdread 2026-10-11

New agents can start from ready made templates, each a working agent for one job. They cover taking a message after hours, updating a customer's details, a post call survey, FAQs, package tracking, caller identification and verification, general customer service, hotel bookings and appointment management. A team picks one when creating the agent and adjusts the flow afterward, adding its own webhook calls, Salesforce steps or Knowledge AI answers where the template stops.

Sourcestudio.docs.ai.vonage.com/ai-studio/templates.mdread 2026-10-11

The customer picks the model for each assistant on the Agent tab. Options include models hosted on Telnyx GPUs, with Kimi as the default, managed third party models from providers such as OpenAI, Gemini and Groq, with OpenAI models on the customer's own key, and any public OpenAI compatible endpoint such as AWS Bedrock, Azure OpenAI, Baseten, vLLM or SGLang. Prompt nodes in a workflow can override the model step by step, and speech to text and text to speech providers are chosen separately.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/custom-llm.mdread 2026-10-11

Knowledge AI runs on Google Gemini models picked by Vonage, and the agentic NLU engine uses language models it does not name. The older Generative AI node let a team connect its own OpenAI key and organization for GPT 3.5 replies, and Vonage is phasing that node out in favor of Knowledge AI. Intent recognition runs on Vonage's own traditional or hybrid NLU engines.

Sourcestudio.docs.ai.vonage.com/voice/nodes/integrations/generative-ai/migrating-from-the-genai-node-to-knowledge-ai.mdread 2026-10-11

The API covers assistants end to end, including creating and updating assistants, assistant tests, canary deploys, MCP server records, SMS chat and a chat endpoint in beta, and the TeXML ai_calls endpoint starts outbound calls. Coding assistants can read the whole reference through llms.txt indexes, and the conversations API hands stored conversations to other systems.

Sourcedevelopers.telnyx.com/development/llms/ai-assistants-llms-txtread 2026-10-11

HTTP agents expose a finished agent as an API. An init call with the account API key and agent ID opens a session and returns a session token, and step calls send the user's input and parameters and return the agent's messages and session status, on regional US and EU hosts. An Insights API returns conversation data under its own authentication, and Vonage's Voice, Messages, SMS and Verify APIs, SDKs and dashboard sit around the studio.

Sourcestudio.docs.ai.vonage.com/http/working-with-http-agents.mdread 2026-10-11

AI tests give each check a name, a target assistant and success criteria, such as what the greeting must say, and a test run shows progress live and reports whether the assistant met every criterion, with the conversation to review. A new version can be saved with notes and sent a share of traffic by ordered rules on the caller's number, all to one version or split by percentage, with the remainder staying on the main version, and rollback clears every rule at once. Coval can generate test conversations from seed cases.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/version-testing-traffic-distribution.mdread 2026-10-11

The tester runs a live conversation by chat or by an outbound call to the builder's phone, with parameter values set in advance to steer the flow, a no input option, a jump from any message to its node on the canvas and a flag on each webhook call showing whether it succeeded, with the raw conversation JSON for failures. Knowledge AI indexes can be tested on their own, showing which source answered. These are hands on checks. Saved test cases, pass or fail results and runs across many conversations at once are not part of the product.

Sourcestudio.docs.ai.vonage.com/ai-studio/chatbot-tester.mdread 2026-10-11

Assistants act through webhooks, MCP servers, integrations and call control tools, and send keypad tones to other phone systems. No hosted browser, remote desktop or screen control is offered.

Sourcedevelopers.telnyx.com/docs/inference/ai-assistants/tools-library.mdread 2026-10-11

Agents reach other systems through webhook, Salesforce, email and custom code nodes, and Knowledge AI reads a single web page when it is added as a source. No hosted browser, remote desktop or screen control is part of AI Studio.

Sourcedeveloper.vonage.com/en/ai-studio/overviewread 2026-10-11