Automating Customer Onboarding in Intercom: Series, Workflows, and Fin AI Explained

Most businesses that adopt Intercom for onboarding end up building something clunky on their first try. Not because the tools are bad, but because Intercom actually gives you three separate systems for automating messages, and it is not obvious which one is meant for which job. If you have ever set up a welcome email in one place, a chat greeting in another, and then wondered why Fin AI never showed up when a new user asked a question mid-sequence, this is why.

Here is how the pieces actually fit together, and how to use them so a new customer's first week feels coordinated instead of like three different systems talking past each other.

The three tools, and what each one is actually for

Series is Intercom's tool for multi-step, time-based campaigns. It is built out of blocks: rule blocks decide who enters and under what conditions, content blocks send the actual message across channels like email, in-app posts, product tours, push notifications, carousels, and banners, wait blocks add delays between steps, and tag blocks update a customer's tags as they move through the journey. Series is the right tool when you want to say "three days after signup, if this person hasn't done X, send this message" without anyone touching it again.

Workflows is Intercom's real-time automation builder. Instead of running on a calendar, Workflows react to events as they happen: a customer opening a conversation, sending a first message, visiting a page, clicking something, a conversation going unanswered, a ticket changing state, or another workflow finishing. Inside a workflow you can branch based on conditions, collect information from the customer, update their attributes, create or close tickets, sync data to outside systems through data connectors, tag conversations, loop in a teammate, and hand a message to Fin AI to answer.

Fin AI Agent is not really a separate automation tool in the same sense. It is the AI layer you plug into Workflows (and into Series and outbound messages) at the point where a real answer is needed instead of a scripted one.

Where each one belongs in an onboarding flow

A well-built onboarding setup usually uses all three, each doing the part it is actually good at.

Series carries the scheduled, one-to-many side of onboarding: the welcome email that goes out the moment someone signs up, a product tour a couple of days later, a check-in message a week in if a key action hasn't happened yet, and light re-engagement nudges further down the line. This is the connective tissue of the onboarding journey, and it does not need real-time intelligence because it is not reacting to a live conversation.

Workflows carry the reactive side. When a new customer opens the chat during onboarding to ask something, a workflow triggered on "customer sends their first message" or "customer opens a new conversation" is what decides what happens next. That is also where a "Let Fin answer" step belongs, so the AI agent picks up the question immediately instead of the customer waiting for a human to notice a new chat.

Fin AI Agent is what actually answers inside that workflow, and it can also follow up after a Series message goes out. If someone replies to an onboarding email or a workflow-driven message with a question, you can add a follow-up action so Fin responds instead of the reply sitting in a queue. For outbound messages this only works with a simple deploy setup, and Fin will give one answer and then hand off to your assigned team regardless of whether the question got fully resolved, so it is meant to catch the easy stuff fast, not to run the entire follow-up conversation on its own.

Two setup details that cause most of the problems

The first is trigger conflicts. If you build a workflow using a broad trigger like "customer sends any message" and also give Fin a step inside it, Fin can end up replying indefinitely, including after a human teammate has already taken over the conversation. Intercom's own guidance is to use narrower triggers like "first message" or "new conversation opened" specifically to avoid this.

The second is treating email like chat. Not every workflow step behaves the same way across channels, so Intercom recommends building separate workflows for email rather than trying to reuse a chat-oriented workflow there. If your onboarding sequence spans email and in-app messaging, plan for that split from the start rather than retrofitting it later.

It is also worth remembering that Fin AI Agent requires a paid plan and that building or editing workflows requires admin permissions, so if you are handing this off to someone on your team, make sure their access actually covers what you are asking them to build.

What good looks like

When this is set up well, a new customer gets a scheduled welcome sequence through Series that does not need daily attention, and the moment they have an actual question, a workflow hands it to Fin AI, which answers using your help center content and escalates to a person only when it should, based on things like a direct request for a human, a frustrated reply, or repeated back and forth without resolution. Nothing about the automation should be visible to the customer as automation. It should just feel like the business is responsive.

Getting there usually means starting smaller than people expect. Pick one onboarding moment, wire it end to end between Series, a workflow, and Fin, watch how real customers actually move through it, and only then expand. Businesses that try to build the whole onboarding journey at once tend to end up with the disconnected mess this article opened with.

If you're setting up or tuning Intercom's Fin AI agent for your business and want help getting the onboarding automation actually coordinated instead of cobbled together, reach out and we can look at what your current setup is missing.

Cover photo by Kelly Sikkema on Unsplash.