# Relevance AI Integrations: Triggers, Plans and App Setup

> How Relevance AI connects to 2,000+ apps: which triggers each plan unlocks, Gmail, Slack, Teams and HubSpot setup, and the webhook route for Make and n8n.

_Source: https://professionalstoolkit.com/articles/relevance-ai-integrations — The Professional's Toolkit · updated November 2026_

---

Relevance AI connects to **2,000+ apps**: native integrations it builds itself, plus Pipedream-based ones that fill out the rest of the catalog [1]. More than 1,000 app triggers work on every plan, and so do webhooks, API calls and tools used as triggers. The plan matters in two places. Premium triggers for WhatsApp, LinkedIn and Telegram start on Pro, the entry plan, while Salesforce, Snowflake and Zendesk triggers need Enterprise, even though their tool steps run on every plan [2]. Make and n8n, like any system that can send an HTTP request, reach an agent through a webhook trigger. For Gmail, Outlook, Slack, Teams, HubSpot and WhatsApp, Pro covers it. Zendesk, Salesforce and Snowflake events are the exception.


**2,000+** — Apps Relevance AI connects to, native plus Pipedream-based integrations



- **1,000+** — App triggers available on every plan
- **From Pro** — Premium WhatsApp, LinkedIn and Telegram triggers
- **Enterprise** — Needed for Salesforce, Snowflake and Zendesk triggers


## How a connection reaches your agent

Native integrations ship with pre-built tool steps and trigger events, and many carry a docs note that Relevance AI built them and supports them itself. That matters when something breaks: for Apollo, Lusha, ZoomInfo, Firmable and Outlook, the support question goes to Relevance AI, not to the app's maker. Pipedream integrations stretch the list to 2,000+ services, and the vendor keeps replacing some with native versions, as it did for Gong, Miro, Content Snare and WorldNewsAPI.

Sign-in comes in two forms. Gmail, Google Calendar, Canva, Zendesk and Miro use OAuth, so you click through a consent screen. Apollo, Lusha, Firmable and Airtop take an API key from your own account with that vendor, while ZoomInfo takes your ZoomInfo email, a client ID and a private key. For supported integrations such as Lusha and Airtop, a key saved on the Integrations page is used by every tool step that needs it. You can connect several accounts of one service, and the docs set no hard limit on how many integrations an organization connects. Without a dedicated integration, a webhook or an API tool step reaches the service instead.

**Data vendors bill separately.** Enrichment through Apollo, Lusha, ZoomInfo or Firmable spends credits on your plan with that vendor, not on Relevance AI's bill. ZoomInfo searches are free, while enrichment draws ZoomInfo credits; once those run out, enrichment fails and search keeps working. On the Relevance AI side, every tool run an agent makes through an integration counts as one Action, failed runs included, and the [tools guide](/articles/relevance-ai-tools) shows how combining steps into one tool saves Actions. If a connection fails, the system tries to reconnect, then notifies you, and the agent keeps its other capabilities.


> 💡 **Give each agent its own inbox:** Several Gmail accounts can be connected to one organization, with different agents using different accounts, so a support agent and a sales agent never read the same mail.


## Which trigger your plan unlocks

| Trigger type | Examples | Plans |
|---|---|---|
| App triggers (1,000+) | Gmail, Outlook, HubSpot, Slack, Microsoft Teams, Freshdesk | Pro, Team, Enterprise |
| Premium triggers | WhatsApp (personal and Business), LinkedIn, Telegram | Pro, Team, Enterprise |
| Build-your-own | Webhook, API, JavaScript SDK, tools as triggers | Pro, Team, Enterprise |
| Enterprise triggers | Salesforce, Snowflake, Zendesk | Enterprise only |

Every organization now needs a paid plan, and Relevance AI's pricing pages describe no free trial, so the first three rows start at Pro, which costs $19 a month billed annually or $29 a month billed monthly for the whole organization. The fourth row is the real decision. Salesforce, Snowflake and Zendesk triggers moved to Enterprise in September 2025, though Salesforce connections set up before then keep working unless removed. Enterprise pricing comes from sales. Below Enterprise, an agent can still write to Zendesk or Salesforce. It just cannot be woken by their events.

Triggers can be paused one at a time or all at once, a control agent triggers gained in March 2026 after Workforces got it in January. Since September 2026, a trigger status page and failure alarms report problems by email, Slack or Teams. Enterprise admins also get integration controls that switch individual integrations and models on or off for the team.

## Gmail, Outlook, Slack and Teams

### Gmail and Outlook

Gmail connects by OAuth with permission to read, send and manage email, and messages your agent sends land in the Gmail Sent folder. A Gmail trigger can fire on new mail, a label, a sender, a subject line or unread mail, at a check frequency you set. Outlook works with Outlook.com, Office 365 and Microsoft 365 accounts. It brings 17 pre-built steps (seven for email, six for calendar, four for contacts) plus an API Call step for Microsoft Graph, which also reaches shared mailboxes through delegated access, with Graph rate limits applying.

Two Outlook details catch people. **The first email can take up to 30 minutes** to arrive after you set the trigger up; later ones usually take a few minutes. Without a search filter, the trigger forwards every message the account receives, and it skips any email with an attachment larger than 25 MB, a Microsoft Graph limit. The docs suggest putting email-sending tools in approval mode first, so the agent asks before each send, and switching to autopilot once its drafts read right.


> ⚠ **One Microsoft connection, four apps:** Removing the Microsoft integration in Relevance AI disables every trigger and tool step on that account across Teams, Outlook, SharePoint and OneDrive.


### Slack

Slack needs two connections. A workspace-level bot, which a Slack workspace admin must approve, powers agents, tools, triggers and Workforces; each person can then connect their own Slack account through OAuth so Relevance AI can act on their behalf. Triggers fire in channels or your own DM, on a keyword or when someone tags the bot, and show live status in the thread. A DM trigger needs Editor rights or higher on that agent, and Exclude Keywords keep a trigger quiet. Use the same email for Slack and Relevance AI, or channels may not appear in the setup. Since January 2026, you can approve or reject an agent's tool requests inside Slack.

### Microsoft Teams

Teams setup has two required steps: connect the Microsoft account by OAuth, then install the Relevance AI app, listed in the Teams App Store since March 2026. The app must be added to every channel or group chat where the agent should answer, and enterprise tenants may need consent from a Global Administrator or Application Administrator. Teams message triggers fire on new messages only, not on new chats or groups, with optional keywords and an Outreach replies only toggle; a separate Teams Calendar trigger starts agents when meetings start or end. Replies arrive as threaded Adaptive Cards that stay active for 30 minutes. One Teams message can start a whole Workforce, though most Teams tool steps are still in beta.

## HubSpot, Salesforce, Zendesk, Freshdesk and Shopify

HubSpot works in the opposite direction from most triggers. **You build the trigger inside HubSpot**: a HubSpot workflow, started by a form fill, an email link click, a lead score being reached or a contact property update, includes an action that sends a message to the connected agent [3]. Going the other way, the HubSpot API Call step lets the agent manage contacts, notes, tasks and engagements, and a HubSpot Assistant in the Marketplace handles contacts, deals and companies in plain English. Campaign setups built on this are in our [marketing guide](/articles/relevance-ai-for-marketing).

On Enterprise, the Salesforce trigger runs an SOQL query that checks every minute unless you change the frequency, with a cadence setting that holds the rest in a queue. The first person to authenticate needs System Administrator rights or the Approve Uninstalled Connected Apps permission. On every plan, the Salesforce API call step can create leads, notes and tasks. Zendesk follows the same split: triggers on new tickets, status changes, priority updates, comments and assignments are Enterprise-only, while ticket steps, including a Delete Ticket step that is permanent, run on every plan.

Freshdesk is not on the Enterprise trigger list. It can start an agent from the agent's Triggers tab once you enter your Freshdesk domain and an API key, and a Freshdesk API Call step opens its full REST API. For stores, Relevance AI's Shopify page lists 53 pre-built tool steps, 35 of them actions, and 18 triggers. Connecting it takes a Shop ID, an Admin API access token and a Storefront API access token.

## WhatsApp, Telegram, Twilio and webhooks

WhatsApp is a premium trigger, available from Pro, for personal and Business accounts; a Business number must be a verified WhatsApp Business account in good standing. Triggers can listen to all incoming messages, chosen contacts or groups, or a content filter. If you connected WhatsApp before mid-March 2026, a provider change means **you must reconnect**, and earlier conversations restart on a new thread. Relevance AI adds no message cap beyond WhatsApp's own policies and your plan's allowance, but bulk sends need recipient opt-ins.

Telegram, also premium, signs in a personal or dedicated account, and the agent talks as that account. BotFather bot tokens are not supported. For a true bot, the docs point to a custom agent with a webhook trigger and API calls. Twilio has no event triggers. Inbound SMS or calls reach an agent only when you point the Twilio number's webhook URL at the agent's webhook trigger, which is the same pattern every unsupported system uses.

### Zapier, Make and n8n

Any system that can send an HTTP request can trigger an agent through a webhook, with a message template, optional Thread ID mapping to keep one conversation together, and an optional Unique ID that ignores a repeated event [4]. That is how Make and n8n connect. Zapier is different. You configure it in Zapier with no setup inside Relevance AI, and its Message Workforce event can start a whole Workforce from any app Zapier supports. The API trigger works as well, but it returns no direct response, so results go to a destination such as Google Sheets or a webhook; our [API guide](/articles/relevance-ai-api) covers keys and regional endpoints.


> 💡 **Let agents listen before replying:** Slack's No Agent Reply option and the No reply method on Teams triggers let an agent process messages without posting back, a quiet way to watch it in a live channel.



## FAQ

**Can LinkedIn activity trigger a Relevance AI agent?**

Yes. LinkedIn is a premium trigger available on the Pro plan and above, alongside WhatsApp and Telegram.

**Which AI model providers can I connect to Relevance AI?**

The docs list OpenAI, Anthropic (also through AWS Bedrock or Vertex AI), Gemini (AI Studio or Vertex), Azure and OpenRouter. Your own model keys are optional on every plan and cut Vendor Credit spend, but not the subscription or Actions.

**Can Relevance AI agents call other MCP servers?**

Yes, through the MCP client, which reaches remote servers over Streamable HTTP and has presets for Notion, Canva and Atlassian. Local MCP servers are not supported.

**Can a calendar event start a Relevance AI agent?**

Yes. Google Calendar connects by OAuth and its trigger can monitor chosen calendars with filters, and since February 2026 a Teams Calendar trigger starts agents when Teams meetings start or end.


## What to wire up first on a Pro plan

Start with the trigger that already carries your work. For most teams that is a Gmail or Outlook inbox, a Slack or Teams channel, or a HubSpot workflow, and each runs on Pro. Put sending tools in approval mode, add search filters before anything goes to autopilot, and map a Unique ID on every webhook so a resent event never runs twice. On the trigger side, only one need forces the Enterprise conversation: an agent woken by Salesforce, Snowflake or Zendesk events. If that is your core trigger, get Enterprise pricing before you build.



## Sources

[1] Relevance AI documentation: Integrations — https://relevanceai.com/docs/integrations/introduction (2026-11)
[2] Relevance AI documentation: Pricing (plans, Actions, Vendor Credits) — https://relevanceai.com/docs/get-started/pricing (2026-11)
[3] Relevance AI documentation: HubSpot — https://relevanceai.com/docs/integrations/popular-integrations/hubspot (2026-11)
[4] Relevance AI documentation: Webhook Triggers — https://relevanceai.com/docs/build/agents/build-your-agent/agent-triggers/custom-webhook (2026-11)
[5] Relevance AI Terms and Conditions — https://relevanceai.com/terms-and-conditions
[6] Relevance AI documentation: Changelog — https://relevanceai.com/docs/changelog