Connect Twilio to your AI agents.

Connect Twilio data to AI agents with Combined. Explore calls, messages, accounts, source setup, MCP access and read-only business queries.

Put Twilio records to work in the agent you already use. Combined syncs business data into a managed data layer and exposes scoped, read-only queries through MCP.

Start with 5 million MAR without a card; usage after the trial is $5 per million MAR, with no monthly minimum. Combined offers 1,125 connectors across catalog sources and connectors generated through desktop automation.

How to connect Twilio to Combined

  1. Add your source. Open Connected Apps in Combined and select Twilio. Use the account or workspace that contains the records you want to analyze.
  2. Authorize access. Complete the connection flow and choose the required permissions. The setup reference below shows the upstream source's configurable fields; the Combined flow may handle some of them for you.
  3. Select datasets and sync. Choose the relevant data, set any available history filters, and wait for the first successful sync.
  4. Grant your agent access. Use the Set up or Access workspace to connect your agent through MCP and scope it to the sources it needs.
  5. Ask a first question. Discover the synced schema, check freshness, and run a bounded read-only query. Then add another source for a cross-app analysis.

The onboarding guide covers the workspace flow. Prepare these Twilio settings before connecting:

Twilio source setup reference
SettingWhen neededInput
Account IDRequired by source schemaCredential entered in the connection flow
Auth TokenRequired by source schemaCredential entered in the connection flow
Lookback windowOptionalinteger
Number of concurrent threadsOptionalinteger
Slice Step DurationOptionalstring
Replication Start DateRequired by source schemastring

Authentication choices are alternatives. Supply the settings for your selected method; enter credentials in the connection flow, rather than in an agent prompt.

Twilio datasets you can evaluate

This reference comes from the upstream source catalog and connector definition, version 1.0.13. Your account permissions, selected streams and successful sync determine what your agent can query. Discover the actual table and field names in Combined before writing SQL.

Twilio dataset and schema reference
DatasetRecord keyUpdate fieldExample fields
callssidInspect sync configurationsid, date_created, date_updated, parent_call_sid, account_sid (27 fields in reference)
messagessidInspect sync configurationaccount_sid, api_version, body, date_created, date_sent (20 fields in reference)
accountssidInspect sync configurationauth_token, date_created, date_updated, friendly_name, owner_account_sid (10 fields in reference)
addressessidInspect sync configurationaccount_sid, city, customer_name, date_created, date_updated (16 fields in reference)
alertssiddate_generatedlog_level, resource_sid, date_updated, account_sid, url (15 fields in reference)
applicationssidInspect sync configurationaccount_sid, api_version, date_created, date_updated, friendly_name (21 fields in reference)
available_phone_number_countriesInspect schemaInspect sync configurationcountry_code, country, uri, beta, subresource_uris
available_phone_numbers_localInspect schemaInspect sync configurationfriendly_name, phone_number, lata, rate_center, latitude (13 fields in reference)
available_phone_numbers_mobileInspect schemaInspect sync configurationfriendly_name, phone_number, lata, rate_center, latitude (13 fields in reference)
available_phone_numbers_toll_freeInspect schemaInspect sync configurationfriendly_name, phone_number, lata, rate_center, latitude (13 fields in reference)
conference_participantsaccount_sid + conference_sidInspect sync configurationaccount_sid, call_sid, label, conference_sid, date_created (14 fields in reference)
conferencessidInspect sync configurationaccount_sid, date_created, date_updated, api_version, friendly_name (12 fields in reference)
conversation_participantssidInspect sync configurationaccount_sid, conversation_sid, sid, identity, attributes (12 fields in reference)
conversation_messagessidInspect sync configurationsid, account_sid, conversation_sid, body, media (15 fields in reference)
conversationssidInspect sync configurationsid, account_sid, chat_service_sid, messaging_service_sid, friendly_name (13 fields in reference)
dependent_phone_numberssidInspect sync configurationsid, account_sid, friendly_name, phone_number, voice_url (26 fields in reference)
executionssidInspect sync configurationsid, account_sid, flow_sid, contact_sid, contact_channel_address (11 fields in reference)
flowssidInspect sync configurationsid, account_sid, date_created, date_updated, friendly_name (9 fields in reference)
incoming_phone_numberssidInspect sync configurationsid, account_sid, friendly_name, phone_number, voice_url (35 fields in reference)
keyssidInspect sync configurationdate_created, date_updated, friendly_name, sid
outgoing_caller_idssidInspect sync configurationaccount_sid, date_created, date_updated, friendly_name, phone_number (7 fields in reference)
message_mediasidInspect sync configurationsid, account_sid, parent_sid, content_type, date_created (7 fields in reference)
queuessidInspect sync configurationaccount_sid, average_wait_time, current_size, date_created, date_updated (10 fields in reference)
recordingssidInspect sync configurationaccount_sid, api_version, call_sid, conference_sid, date_created (19 fields in reference)
rolessidInspect sync configurationsid, account_sid, chat_service_sid, friendly_name, type (9 fields in reference)
servicessidInspect sync configurationsid, account_sid, friendly_name, date_created, date_updated (7 fields in reference)
stepsidInspect sync configurationparent_step_sid, name, date_updated, transitioned_to, account_sid (13 fields in reference)
trunkssidInspect sync configurationsid, account_sid, domain_name, disaster_recovery_method, disaster_recovery_url (17 fields in reference)
transcriptionssidInspect sync configurationaccount_sid, api_version, date_created, date_updated, duration (13 fields in reference)
usage_recordsaccount_sid + category + start_date + end_dateInspect sync configurationcategory, count, price_unit, subresource_uris, description (15 fields in reference)
usage_triggerssidInspect sync configurationusage_record_uri, date_updated, date_fired, friendly_name, uri (16 fields in reference)
user_conversationsaccount_sidInspect sync configurationnotification_level, unique_name, user_sid, friendly_name, conversation_sid (18 fields in reference)
userssidInspect sync configurationis_notifiable, date_updated, is_online, friendly_name, account_sid (13 fields in reference)
verify_servicessidInspect sync configurationdefault_template_sid, tts_name, psd2_enabled, do_not_share_warning_enabled, mailer_sid (19 fields in reference)

A cursor tracks changes for a source stream. It is not automatically the right date for a business metric: invoice date, close date and record-update date can answer different questions.

Useful Twilio questions for your AI agent

Start with the datasets you have granted. These prompts turn the referenced schema into checks and summaries; the agent should verify each field against your synced data.

  • Count distinct calls using sid and check for duplicate keys.
  • Count distinct messages using sid and check for duplicate keys.
  • Break down accounts by status and show the largest groups, with a total across all matching records.
  • Count distinct addresses using sid and check for duplicate keys.

Combine Twilio with another business app

Connect the second source, identify a maintained customer, account or project mapping, and aggregate each dataset before joining. Keep unmatched records visible and define the reporting window. The CRM-and-billing worked example includes runnable SQL demonstrating how a many-to-many join can double the result.

Use Twilio data in Claude Code, Cursor or another MCP client

Once the source is synced, the agent connects to Combined's MCP interface. Use the generated account-specific configuration from your workspace, keep credentials local, and grant only the sources needed for the task. Follow the Claude Code walkthrough or the MCP setup reference.

Use Combined to analyze my Twilio data.
List my granted sources, discover the datasets and describe their fields.
Look for datasets corresponding to calls, messages, accounts, addresses, alerts, applications.
Check the last successful sync and report the available date coverage.
Propose three useful questions based on the fields actually available.
For the question I choose, run read-only SQL over the complete matching dataset.
Use a bounded result; include source names, time window and query receipt.
Ask for a missing join key or metric definition instead of inventing one.

Download this Twilio guide as Markdown to use alongside your agent instructions.

Which Twilio connection fits the task?

Managed data for complete questions across apps

Choose Combined when you want to analyze synced records, calculate totals, or join Twilio with other business data without operating a separate ingestion and query stack. The agent works with your source grants and read-only SQL access.

Application tools for live operations

For a task that must change Twilio immediately, evaluate its available API, native agent tools or desktop workflow. Keep application-action permissions separate from read-only analytics access. A live lookup and an aggregate over synced history solve different jobs.

Desktop-generated connectors for workflows beyond the catalog

Combined's 1,125-connector coverage includes connections generated through computer use. Choose this route when the workflow depends on an application's desktop or browser interface, and establish the fields, access and refresh behavior for that connection.

Read the direct MCP versus managed data comparison and compare business-data MCP options.

Troubleshoot your Twilio connection

  • Authorization fails: check the selected authentication method, credential validity, source permissions and target account.
  • History is missing: check the source's start-date or history filter and allow the initial sync to finish before comparing totals.
  • The wrong records appear: verify the workspace, account, project or domain selected in the source settings.
  • A dataset is absent: check whether it is enabled and permitted for your account, then inspect discovery and sync status.
  • The agent cannot query it: confirm the source is included in the agent's grant and that its sync succeeded.
  • A total looks wrong: inspect record keys, duplicate rows, time zones, currencies and join fanout before changing the prompt.

Source details reviewed September 10, 2026. Twilio upstream connector documentation · Versioned connector definition. See also Combined Sources and data freshness for AI agents.