Retell AI connector for AI agents

Tools your voice ops agent reaches for on Retell AI, scoped per user.

CALL ANY TOOL

83 tools for voice agents in Retell AI: create, update, and publish voice and chat agents, place phone and batch calls, read calls and chats, manage contacts and phone numbers, and run batch tests.

retell_add_community_voice

Add community voice

Add a community voice from a voice provider to the Retell workspace's voice library. Returns the voice object, including voice_id, voice_name, provider, accent, gender, age and preview_audio_url. Use retell_add_community_voice after finding a voice with retell_search_community_voice; then use the returned voice_id in retell_create_agent.

Parameters

Name

Type

Required

Description

provider_voice_id

string

Required

Voice ID assigned by the provider, from retell_search_community_voice. Example: 21m00Tcm4TlvDq8ikWAM

voice_name

string

Required

Name to give the voice in your library. Example: Support Voice

public_user_id

string

Optional

User ID of the voice owner. Required for ElevenLabs voices only. Example: 63e06b7e7cafdc46be4d2e0b

voice_provider

string

Optional

Provider the voice comes from: elevenlabs, cartesia, minimax, fish_audio or inworld. Example: elevenlabs. One of: `elevenlabs`, `cartesia`, `minimax`, `fish_audio`, `inworld`.

retell_agent_playground_completion

Agent playground completion

retell_create_agent

Create agent

Create a Retell voice agent from a response engine and a voice, with optional language, voice, timing, webhook and post-call analysis settings. Returns the agent object, including agent_id, version, is_published, response_engine, agent_name, voice_id, language and the voice, webhook and call settings. Use retell_create_agent for phone and web call agents. Use retell_create_chat_agent for chat and SMS agents. Create the response engine first with retell_create_retell_llm or retell_create_conversation_flow, and pick a voice with retell_list_voices.

Parameters

Name

Type

Required

Description

response_engine

object

Required

Response engine that generates the agent's replies, as an object with a type. Use {"type": "retell-llm", "llm_id": "<llm_id>"} for a Retell LLM (see retell_create_retell_llm), {"type": "conversation-flow", "conversation_flow_id": "<id>"} for a conversation flow, or {"type": "custom-llm", "llm_websocket_url": "wss://..."} for your own LLM server. Retell LLM and conversation flow engines may also take a numeric version. Example: {"type": "retell-llm", "llm_id": "llm_234sdertfsdsfsdf"}

voice_id

string

Required

ID of the voice the agent speaks with, from the `voice_id` field returned by retell_list_voices. Example: retell-Cimo

agent_name

string

Optional

Name of the agent, for your own reference. Example: Customer Support Agent

ambient_sound

string

Optional

Background sound mixed into the call to sound more natural: coffee-shop, convention-hall, summer-outdoor, mountain-outdoor, static-noise or call-center. Leave empty for none. One of: `coffee-shop`, `convention-hall`, `summer-outdoor`, `mountain-outdoor`, `static-noise`, `call-center`.

begin_message_delay_ms

integer

Optional

Milliseconds to wait before the agent's first message, from 0 to 5000. Example: 1000

boosted_keywords

array

Optional

Words the transcriber should favour, such as names and product terms. Example: ["Retell", "Kubernetes"]

data_storage_setting

string

Optional

What Retell stores for this agent's calls: everything, everything_except_pii, or basic_attributes_only (metadata only, no transcripts or recordings). One of: `everything`, `everything_except_pii`, `basic_attributes_only`.

enable_backchannel

boolean

Optional

Whether the agent interjects short acknowledgements such as "yeah" or "uh-huh" while the user speaks. Defaults to false.

end_call_after_silence_ms

integer

Optional

End the call when the user stays silent this many milliseconds after the agent speaks. Minimum 10000; defaults to 600000 (10 minutes). Example: 30000

interruption_sensitivity

number

Optional

How easily the user can interrupt the agent, from 0 (cannot interrupt) to 1 (interrupts easily). Defaults to 1. Example: 0.7

language

string

Optional

Locale the agent speaks and transcribes, such as en-US, es-ES or de-DE. Defaults to en-US. Example: en-US. One of: `en-US`, `en-IN`, `en-GB`, `en-AU`, `en-NZ`, `de-DE`, `es-ES`, `es-419`, `hi-IN`, `fr-FR`, `fr-CA`, `ja-JP`, `pt-PT`, `pt-BR`, `zh-CN`, `ru-RU`, `it-IT`, `ko-KR`, `nl-NL`, `nl-BE`, `pl-PL`, `tr-TR`, `vi-VN`, `ro-RO`, `bg-BG`, `ca-ES`, `th-TH`, `da-DK`, `fi-FI`, `el-GR`, `hu-HU`, `id-ID`, `no-NO`, `sk-SK`, `sv-SE`, `lt-LT`, `lv-LV`, `cs-CZ`, `ms-MY`, `af-ZA`, `ar-SA`, `az-AZ`, `bs-BA`, `cy-GB`, `fa-IR`, `fil-PH`, `gl-ES`, `he-IL`, `hr-HR`, `hy-AM`, `is-IS`, `kk-KZ`, `kn-IN`, `mk-MK`, `mr-IN`, `ne-NP`, `sl-SI`, `sr-RS`, `sw-KE`, `ta-IN`, `ur-IN`, `yue-CN`, `uk-UA`.

max_call_duration_ms

integer

Optional

Maximum call length in milliseconds, from 60000 (1 minute) to 7200000 (2 hours). Defaults to 3600000 (1 hour). Example: 1800000

post_call_analysis_data

array

Optional

Extra data to extract from each call after it ends. Each item has a type (string, enum, boolean or number), a name and a description; enum items also list choices. Example: [{"type": "string", "name": "order_id", "description": "The order ID the caller mentions."}]

responsiveness

number

Optional

How quickly the agent responds, from 0 (waits longer) to 1 (responds fastest). Defaults to 1. Example: 0.8

timezone

string

Optional

IANA timezone for the agent, used for time-aware prompts. Defaults to America/Los_Angeles. Example: America/New_York

voice_model

string

Optional

Voice model for the selected voice, when its provider offers several (for example eleven_flash_v2_5, eleven_multilingual_v2 or sonic-3). Leave empty to use the voice's default model.

voice_speed

number

Optional

Speech rate from 0.5 (slower) to 2 (faster). Defaults to 1. Example: 1.1

voice_temperature

number

Optional

Voice stability from 0 (more stable) to 2 (more varied). Defaults to 1. Example: 1

voicemail_option

object

Optional

What to do when the call reaches voicemail in the first 3 minutes. The action is an object with a type: hangup, static_text (with text to speak), prompt (with text describing what to say) or bridge_transfer. Example: {"action": {"type": "static_text", "text": "Please call us back at 555-0100."}}

webhook_url

string

Optional

URL that receives this agent's call event webhooks (call_started, call_ended, call_analyzed by default). Example: https://example.com/retell/webhook

retell_create_batch_test

Create batch test

retell_create_chat_completion

Create chat completion

retell_create_conversation_flow_component

Create conversation flow component

retell_create_retell_llm

Create retell llm

Create a Retell LLM response engine: the model, prompt, opening message, tools and optional states that drive an agent's replies. Returns the Retell LLM object, including llm_id, version, is_published, model, general_prompt, begin_message, general_tools, states and knowledge_base_ids. Use retell_create_retell_llm for prompt-based agents. Use retell_create_conversation_flow for node-based flows. Attach it to an agent with retell_create_agent or retell_update_agent using response_engine {"type": "retell-llm", "llm_id": ...}.

Parameters

Name

Type

Required

Description

start_speaker

string

Required

Who speaks first in the conversation: user or agent. The Retell docs mark it required when creating a Retell LLM. Example: agent. One of: `user`, `agent`.

begin_message

string

Optional

First thing the agent says when it speaks first. Leave empty to let the LLM generate the opening. Example: Hi, thanks for calling Acme. How can I help?

default_dynamic_variables

object

Optional

Default values for dynamic variables used in the prompt, as string key-value pairs; per-call values override them. Example: {"company_name": "Acme"}

general_prompt

string

Optional

System prompt applied in every state of the conversation: the agent's role, tone and instructions. Example: You are a friendly receptionist for Acme Dental. Book, move or cancel appointments.

general_tools

array

Optional

Tools the LLM can call in any state. Each tool has a type and a name; types include end_call, transfer_call, press_digit, send_sms, custom (calls your URL), code, extract_dynamic_variable, agent_swap, mcp and integration_app, each with its own extra fields. Example: [{"type": "end_call", "name": "end_call", "description": "End the call when the user says goodbye."}]

knowledge_base_ids

array

Optional

IDs of knowledge bases the LLM can retrieve from, from retell_list_knowledge_bases. Example: ["knowledge_base_a456426614174000"]

model

string

Optional

Text LLM that powers the agent, such as gpt-4.1, gpt-4.1-mini or gpt-5. See the Retell docs for the current list of supported models. Example: gpt-4.1

model_temperature

number

Optional

Randomness of responses, from 0 (deterministic) to 1 (more varied). Defaults to 0. Example: 0.2

s2s_model

string

Optional

Speech-to-speech model to use instead of a text model, such as gpt-realtime. Set either model or s2s_model, not both. Example: gpt-realtime

starting_state

string

Optional

Name of the state the conversation starts in. Required when states is not empty. Example: greeting

states

array

Optional

Optional conversation states. Each state has a unique name, a state_prompt appended to the general prompt, edges to other states and state-specific tools. Requires starting_state. Example: [{"name": "greeting", "state_prompt": "Greet the caller and ask how you can help."}]

retell_delete_agent

Delete agent

retell_delete_call

Call delete

retell_delete_contact

Delete contact

retell_delete_conversation_flow_component

Delete conversation flow component

retell_delete_phone_number

Delete phone number

retell_get_agent

Get agent

retell_get_call

Call get

retell_get_contact

Get contact

retell_get_conversation_flow_component

Get conversation flow component

retell_get_retell_llm

Get retell llm

retell_list_agent_versions

List agent versions

retell_list_calls

List calls

retell_list_contacts

List contacts

retell_list_export_requests

List export requests

retell_list_retell_llms

List retell llms

retell_publish_agent_version

Publish agent version

retell_repair_agent

Repair agent

retell_search_community_voice

Search community voice

retell_stop_call

Call stop

retell_update_agent

Update agent

Update settings on a Retell voice agent's latest draft version; fields you leave out stay unchanged. Returns the agent object, including agent_id, version, is_published, response_engine, agent_name, voice_id, language and the voice, webhook and call settings. Use retell_update_agent to change voice, timing or webhook settings. Use retell_update_retell_llm or retell_update_conversation_flow to change the prompt. Changes apply to the draft; publish them with retell_publish_agent_version.

Parameters

Name

Type

Required

Description

agent_id

string

Required

ID of the voice agent, from the `agent_id` field returned by retell_list_agents. Example: oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD

agent_name

string

Optional

Name of the agent, for your own reference. Example: Customer Support Agent

ambient_sound

string

Optional

begin_message_delay_ms

integer

Optional

Milliseconds to wait before the agent's first message, from 0 to 5000. Example: 1000

boosted_keywords

array

Optional

Words the transcriber should favour, such as names and product terms. Example: ["Retell", "Kubernetes"]

data_storage_setting

string

Optional

enable_backchannel

boolean

Optional

Whether the agent interjects short acknowledgements such as "yeah" or "uh-huh" while the user speaks. Defaults to false.

end_call_after_silence_ms

integer

Optional

End the call when the user stays silent this many milliseconds after the agent speaks. Minimum 10000; defaults to 600000 (10 minutes). Example: 30000

interruption_sensitivity

number

Optional

How easily the user can interrupt the agent, from 0 (cannot interrupt) to 1 (interrupts easily). Defaults to 1. Example: 0.7

language

string

Optional

max_call_duration_ms

integer

Optional

Maximum call length in milliseconds, from 60000 (1 minute) to 7200000 (2 hours). Defaults to 3600000 (1 hour). Example: 1800000

post_call_analysis_data

array

Optional

response_engine

object

Optional

responsiveness

number

Optional

How quickly the agent responds, from 0 (waits longer) to 1 (responds fastest). Defaults to 1. Example: 0.8

timezone

string

Optional

IANA timezone for the agent, used for time-aware prompts. Defaults to America/Los_Angeles. Example: America/New_York

version

integer

Optional

Draft version to update. Omit for the latest version. Published versions cannot be changed. Example: 3

voice_id

string

Optional

ID of the voice the agent speaks with, from the `voice_id` field returned by retell_list_voices. Example: retell-Cimo

voice_model

string

Optional

Voice model for the selected voice, when its provider offers several (for example eleven_flash_v2_5, eleven_multilingual_v2 or sonic-3). Leave empty to use the voice's default model.

voice_speed

number

Optional

Speech rate from 0.5 (slower) to 2 (faster). Defaults to 1. Example: 1.1

voice_temperature

number

Optional

Voice stability from 0 (more stable) to 2 (more varied). Defaults to 1. Example: 1

voicemail_option

object

Optional

webhook_url

string

Optional

URL that receives this agent's call event webhooks (call_started, call_ended, call_analyzed by default). Example: https://example.com/retell/webhook

retell_update_chat

Update chat

retell_update_conversation_flow

Update conversation flow

Update a Retell conversation flow's model, nodes, tools, prompt or other settings; fields you leave out stay unchanged. Returns the conversation flow object, including conversation_flow_id, version, model_choice, start_speaker, global_prompt, nodes, tools, components and knowledge_base_ids. Use retell_update_conversation_flow to change a flow-based agent's behaviour. Arrays such as nodes replace the stored value, so send the complete list from retell_get_conversation_flow.

Parameters

Name

Type

Required

Description

conversation_flow_id

string

Required

ID of the conversation flow, from the `conversation_flow_id` field returned by retell_list_conversation_flows. Example: a1b2c3d4e5f60718293a4b5c6d7e8f90

begin_after_user_silence_ms

integer

Optional

When the user speaks first, start talking after the user has been silent this many milliseconds. Omit to wait indefinitely. Example: 2000

default_dynamic_variables

object

Optional

Default values for dynamic variables referenced in the flow, as string key-value pairs. Example: {"company_name": "Retell Inc", "support_hours": "9 AM - 5 PM"}

global_prompt

string

Optional

Prompt applied in every node of the flow, such as the agent's role and tone. Example: You are a helpful customer service agent.

knowledge_base_ids

array

Optional

IDs of knowledge bases the flow can retrieve from, from retell_list_knowledge_bases. Example: ["knowledge_base_a456426614174000"]

model_choice

object

Optional

Model that runs the flow. For a text model use {"type": "cascading", "model": "gpt-4.1"} (optionally with high_priority); for speech-to-speech use {"type": "s2s", "model": "gpt-realtime"}. Example: {"type": "cascading", "model": "gpt-4.1"}

model_temperature

number

Optional

Randomness of responses, from 0 (deterministic) to 1 (more varied). Example: 0.7

nodes

array

Optional

Nodes of the flow. Each node has an id, a type and type-specific fields; types include conversation (with an instruction prompt or static text), function, code, transfer_call, press_digit, branch, sms, extract_dynamic_variables, agent_swap, mcp, component and end. Nodes link to each other through edges with a transition_condition and destination_node_id. Example: [{"id": "start", "type": "conversation", "instruction": {"type": "prompt", "text": "Greet the customer and ask how you can help."}}]

start_node_id

string

Optional

ID of the node the conversation starts at. Example: start

start_speaker

string

Optional

Who speaks first in the conversation: user or agent. Example: agent. One of: `user`, `agent`.

tools

array

Optional

Custom tools nodes can call. Each tool has a type (such as custom for an HTTP call), a name, a description and type-specific fields such as url and method. Example: [{"type": "custom", "name": "get_customer_info", "description": "Get customer information", "tool_id": "tool_001", "url": "https://api.example.com/customer", "method": "GET"}]

version

integer

Optional

Version to update. Omit for the latest version. Example: 1

retell_update_phone_number

Update phone number

Update a Retell phone number's bound voice and SMS agents, nickname, webhooks, allowed countries or SIP trunk settings; fields you leave out stay unchanged. Returns the phone number object, including phone_number, phone_number_type, nickname, area_code, inbound_agents, outbound_agents, SMS agents, webhook URLs and allowed country lists. Use retell_update_phone_number to change which agent handles a number's calls or texts.

Parameters

Name

Type

Required

Description

phone_number

string

Required

The phone number in E.164 format (+ country code, then the number with no spaces or symbols), as returned by retell_list_phone_numbers. Example: +14157774444

allowed_inbound_country_list

array

Optional

ISO 3166-1 alpha-2 country codes from which inbound calls are allowed. Empty or unset allows all countries. Example: ["US", "CA"]

allowed_outbound_country_list

array

Optional

ISO 3166-1 alpha-2 country codes to which outbound calls are allowed. Empty or unset allows all countries. Example: ["US", "CA"]

auth_password

string

Optional

New SIP trunk authentication password, for imported numbers.

auth_username

string

Optional

New SIP trunk authentication username, for imported numbers. Example: retell-trunk

fallback_number

string

Optional

Number in E.164 format to forward inbound calls to when concurrency is full. Example: +14155550100

inbound_agents

array

Optional

Agents that answer inbound calls bound to this number, each with an agent_id and a weight (weights must add up to 1); one agent is picked at random by weight for each conversation. Each entry may also pin an agent_version. Pass an empty array to unbind. Example: [{"agent_id": "oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD", "weight": 1}]

inbound_sms_agents

array

Optional

Chat agents that answer inbound SMS bound to this number, each with an agent_id and a weight (weights must add up to 1); one agent is picked at random by weight for each conversation. Each entry may also pin an agent_version. Pass an empty array to unbind. Example: [{"agent_id": "oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD", "weight": 1}]

inbound_sms_webhook_url

string

Optional

URL Retell calls on each inbound SMS so you can override the agent or set dynamic variables. Example: https://example.com/inbound-sms-webhook

inbound_webhook_url

string

Optional

URL Retell calls on each inbound call so you can override the agent or set dynamic variables. Example: https://example.com/inbound-webhook

nickname

string

Optional

Nickname for the number, for your own reference. Example: Support line

outbound_agents

array

Optional

Agents that place outbound calls bound to this number, each with an agent_id and a weight (weights must add up to 1); one agent is picked at random by weight for each conversation. Each entry may also pin an agent_version. Pass an empty array to unbind. Example: [{"agent_id": "oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD", "weight": 1}]

outbound_sms_agents

array

Optional

Chat agents that send outbound SMS bound to this number, each with an agent_id and a weight (weights must add up to 1); one agent is picked at random by weight for each conversation. Each entry may also pin an agent_version. Pass an empty array to unbind. Example: [{"agent_id": "oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD", "weight": 1}]

termination_uri

string

Optional

New SIP trunk termination URI, for imported numbers. Example: someuri.pstn.twilio.com

transport

string

Optional

Outbound SIP transport protocol: TLS, TCP or UDP. One of: `TLS`, `TCP`, `UDP`.

For more tools, view docs.

Build your Agent

Same auth pattern across LangChain, OpenAI, Anthropic, and Google ADK.

import { ScalekitClient } from "@scalekit-sdk/node";
import { createReactAgent } from "@langchain/langgraph/prebuilt";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

// Retell AI tools scoped to this user
const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["retell"], toolNames: [
    "retell_list_calls",
    "retell_get_call",
    "retell_create_phone_call"] },
  pageSize: 100,
});

const agent = createReactAgent({ llm, tools });
await agent.invoke({ messages: [{ role: "user", content: "Find yesterday's negative support calls and call each customer back" }] });

import OpenAI from "openai";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const openai = new OpenAI();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["retell"] }, pageSize: 100,
});

const res = await openai.chat.completions.create({
  model: "gpt-5",
  messages: [{ role: "user", content: "Find yesterday's negative support calls and call each customer back" }],
  tools,
});

// Execute the tool call with the user's vaulted Retell AI access token
await sk.tools.executeTool(res.choices[0].message.tool_calls[0], "user_123");

import Anthropic from "@anthropic-ai/sdk";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const anthropic = new Anthropic();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["retell"] }, pageSize: 100,
});

const msg = await anthropic.messages.create({
  model: "claude-sonnet-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Find yesterday's negative support calls and call each customer back" }],
  tools,
});

// Tool call runs with the user's vaulted Retell AI access token
await sk.tools.executeTool(msg.content, "user_123");

import { Agent } from "@google/adk/agents";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["retell"] }, pageSize: 100,
});

const agent = new Agent({
  name: "voice_ops_agent",
  model: "gemini-2.5-pro",
  instruction: "Retell AI tools scoped to this user",
  tools,
});

await agent.run("Find yesterday's negative support calls and call each customer back");

Try these prompts

Copy any prompt into your agent. Each maps directly to a Retell AI tool. Click to copy, paste into your agent, done.

Agents and flows

Copy the prompt

Copied

List every voice and chat agent in our workspace and whether each one is published.

Copy the prompt

Copied

Update the support agent's draft to use the new voice.

Copy the prompt

Copied

Publish the latest draft version of the booking agent.

Calls and chats

Copy the prompt

Copied

Call +1 415 555 0134 from our sales number with the follow-up agent.

Copy the prompt

Copied

Schedule a batch of reminder calls to these 40 numbers tomorrow at 10am.

Copy the prompt

Copied

Show the transcript and call analysis for the last call from +1 212 555 0198.

Testing and numbers

Copy the prompt

Copied

Run every test case against the intake conversation flow.

Copy the prompt

Copied

Which test runs failed in the last batch test, and why?

Copy the prompt

Copied

List our phone numbers and the agents bound to each one.

SEE HOW AUTH WORKS

Each user connects their own Retell AI access token once, and Scalekit sends it with every call. Access tokens stay vaulted, every call is scope checked, and every action is logged.

Authorize

Your user connects

Retell AI

once. We tie it to their identity and the meetings they approved — no shared bot account, no org-wide access

Who:

user ‘A’

when:

Once per user

access:

Limited to user

Store

Their

Retell AI

token lives in a vault scoped to them. User A's meetings are never reachable by an agent acting for user B, even on the same connection

vault:

encrypted

scope:

per-user

tokens:

auto-refreshed

Resolve

When your agent calls a

Retell AI

tool, we fetch the right token server-side. It never touches your agent, never appears in the LLM context, never shows up in your logs

speed:

~40ms

check:

before every call

seen by:

nobody

Audit

Every

Retell AI

tool call is logged — who triggered it, which meeting was fetched, what came back. 90 days of history, tied to the user who authorized it

history:

90 days

export:

SIEM-ready

logged:

every call

Test other agents

See the same per-user auth pattern across other AI and communication connectors.

Support and Ops Teams

Support ticket automation agent

Fetches new Zendesk tickets, drafts a reply from Notion knowledge base articles, digests what it cannot answer to Slack, and archives the rest, acting as the support agent rather than a shared API key.

GTM and RevOps Teams

Outbound prospecting agent

Searches Apollo for prospects matching your ICP, scores and ranks them, drafts personalized outreach in Gmail, and logs every send to Google Sheets. Mail goes out as the rep, not from a shared inbox.

Why Scalekit

Secure your agent's access. Connectors ship in minutes

01.

Shared tokens break per-user analytics

A shared Retell AI access token looks fine in a demo. In production every outbound call and agent change looks like one service account, and you cannot tell which user triggered it. Scalekit resolves the credential of the actual user who triggered the agent, never a shared bot.

// shared key
audit → bot_service_account

// scalekit
audit → user_abc ✓

02.

Authentication is not authorization

03.

Multi-tenancy is architectural

04.

Retell AI today. Ten connectors tomorrow.

“Our agents act across Salesforce, Gong, Google Drive, and more, on behalf of every customer. Scalekit behind the scenes meant we can keep adding tools without ever rebuilding how credentials or tool calling work.”

Venu Madhav Kattagoni

Head of Engineering / Von

FAQs

Frequently Asked Questions

Does the agent access Retell AI as the user or through a shared access token?

As the user. Each user connects their own Retell AI access token once, and Scalekit sends it with every call. Audit logs attribute every call placed and every agent change to that user, not a shared service account.

Where is the Retell AI access token stored?

In Scalekit's managed AES-256 token vault, namespaced per tenant. To rotate it, the user reconnects with a new access token. Revocation is a single dashboard action. Credentials never appear in prompts, logs, or LLM context.

Can I limit what the agent does in Retell AI?

Yes. Filter by tool name in listScopedTools to expose only what you want, for example the call, chat, and agent read tools without retell_delete_agent, which permanently deletes a voice agent and all of its versions. Scalekit also enforces scope checks before every API call.

What happens when a user revokes Retell AI access?

The connection is invalidated on the next tool call. Subsequent requests for that user fail closed with a clear error. Other users in the tenant remain unaffected. The event is logged for audit.

Can the agent place real phone calls?

Yes. retell_create_phone_call places an outbound call from a Retell number to a destination number, and retell_create_batch_call schedules a batch of outbound calls from one number. retell_stop_call hangs up a call in progress. Expose these tools only to agents that should dial out.