How to Automate Sequences & Follow-Ups in Front with AI (2026)

How Front sequences and follow-ups work

A sequence in Front is a series of messages sent to a list of recipients, with automatic follow-ups for anyone who goes quiet. Per Front's guide to understanding sequences, a sequence can range from 1 to 10 stages (labeled A through J), and each stage after the first is a follow-up to the one before it.

The follow-up logic is the whole point. Each stage has a sending delay that triggers only if the recipient hasn't replied within a set number of days. Front's own example is "Follow up in 7 days if no reply" — so a three-week nudge is simply stages B, C, and D each set to send seven days after the previous one, provided the person still hasn't answered.

Two behaviors make sequences safe to run at scale:

• Stop-on-reply. Front is explicit: "If a recipient sends a reply to a sequence stage, subsequent stages will not send out to their email address." The moment someone engages, the chasing stops for them — you don't badger a prospect who already wrote back. (Automatic out-of-office replies are the sensible exception; they don't stop the sequence.)
• Mail merge personalization. You can customize each message with variables using {{variable}} syntax — pulling CSV columns through {{sequence...}} and contact custom fields through {{recipient...}}, with a fallback like {{variable|'there'}} for missing data. A sequence supports up to 200 recipients and up to 20 custom columns per send.

That's a genuinely capable outbound engine, and for straightforward campaigns it's all most teams need.

The playbook: building a follow-up sequence

Here's the standard flow for a support, sales, or CS follow-up sequence in Front:

1. Build your list. Prepare a CSV whose first column is email, plus up to 20 custom columns (first name, account, renewal date, open ticket ID — whatever you'll merge in).
2. Write stage A. The opener. Use {{recipient.first_name}} and other variables so the "mass" email reads like a one-to-one note. Keep the subject deliberate — Front locks the subject after stage A and prepends "Re:" to later stages.
3. Set the follow-up cadence. For stages B onward, choose the delay: follow up in N days if no reply. A common three-touch pattern is B at +3 days, C at +7, D at +14.
4. Let stop-on-reply do its job. Anyone who replies drops out of the remaining stages automatically. You only keep chasing the silent ones.
5. Review engagement. After sending, check who opened and who replied, and at which stage — then fold that into your next campaign.

The friction is that steps 2 and 3 are the same for everyone. Mail merge swaps in a name and an account, but the argument, the tone, and the timing are fixed at authoring time. A prospect who half-replied "not now" gets the same stage C as one who ghosted entirely. That's the seam where AI earns its place.

Above: a Macha demo-org agent's Schedules — recurring and triggered runs that power automated follow-ups and checks. This is a Macha agent demonstrating the capability, not a live Front conversation.

Where native Front AI already helps

Front has invested heavily in AI, and it's fair to credit what it does well before reaching for anything else. Front's AI features cover three that touch outbound and follow-up work directly:

• AI Compose polishes a draft as you write it — one click to make a message more professional or friendlier, fix typos, or change the length. It's included in the base plans (up to 200 Compose actions per teammate per day), and it's genuinely useful when you're hand-writing a follow-up after a sequence flags a reply.
• Autopilot is Front's AI agent that can resolve conversations end-to-end and reply automatically to conversations tagged with selected topics — so when a sequence reply lands and it's a simple, known question, Autopilot can handle it without a human.
• Smart QA reviews tickets automatically and scores service quality, which is helpful for auditing how your team handles the replies your sequences generate.

One honest caveat worth naming: Front's older AI Answers chatbot is now a legacy feature — "no longer available for purchase" — and Front steers new buyers to Autopilot instead. If you're planning around Front AI today, plan around Autopilot, not AI Answers.

So native Front AI already helps you write a follow-up (Compose) and auto-answer the simple replies (Autopilot). Where it stops is the personalization and reasoning inside the sequence itself.

Where an AI agent layer beats a static sequence

A native sequence is deterministic: fixed stages, fixed copy, fixed cadence. AI Compose helps a human draft, and Autopilot can auto-reply to tagged topics — but neither reaches into the sequence to make each step smarter for each person. That's the job of a dedicated AI agent layer running on top of Front.

Macha is that layer. It runs on top of the Front you already use through the Macha–Front connector — it does not replace Front, your shared inboxes, or your sequences. Here's where an agent adds what a static sequence can't:

• Personalized sequence steps. Instead of one templated stage C for everyone, an agent reads what actually happened — the prospect's last reply, the account's open tickets, the renewal date pulled live from your system — and drafts a follow-up tailored to that situation. Mail merge fills a blank; an agent writes the sentence.
• Smart stop-on-reply. Front's stop-on-reply is binary: any reply halts the chase. An agent can read the reply's intent — "send me pricing" (advance and answer), "wrong person" (route to the right contact), "not interested" (suppress) — instead of treating every response identically.
• Follow-up drafting on a schedule. Macha's Schedules run recurring and triggered agent runs — every hour, every four hours, every day — so an agent can check for unreplied threads, draft context-aware nudges, and surface them for a teammate to send, rather than firing the same static stage on a fixed clock.

The mechanism underneath is that the agent understands intent, not just a merge field, and can pull real data through a custom tool that turns your REST API into something the agent can call at send time. If you want to build one, building an AI agent for Front walks through it, and connecting Front to Macha to route conversations to AI covers the wiring.

The cost model: per conversation vs per action

The two approaches also price differently, and it's worth being clear-eyed about it. Front's plans run $25/seat/mo (Starter), $65/seat/mo (Professional), and $105/seat/mo (Enterprise) per Front's pricing page, with sequences included in the seat. AI capabilities are add-ons: AI Compose ships in the base plans (200 actions/teammate/day), Smart QA is a $20/seat/mo add-on (or included in Enterprise), and Autopilot is billed for its AI work — the pricing page lists it "starting at $0.05/conversation," while Front's Autopilot Resolve documentation describes billing per Resolution (a complete AI response with no human intervention; drafts don't count). Front's pricing tiers lay out exactly what unlocks where.

Macha prices on a different unit. Its credits are consumed per AI action — each thing the agent actually does, such as drafting a follow-up, reading a reply's intent, or calling a custom tool to fetch a renewal date — never per resolution or per deflection. That distinction matters: an automation/orchestration layer does variable amounts of reasoning per conversation, and pricing per action is the honest way to reflect that. Some replies are one action; a multi-step follow-up that pulls data and drafts is several.

The clean division of labour: let Front sequences be the reliable dispatcher that sends and chases, and layer an agent on top for the reasoning — personalizing the step, reading the reply, and drafting the answer. For the mechanics of sequences themselves, see Front sequences explained; for the automation engine that pairs with them, Front rules explained; and for the wider picture, Front AI explained and the category of AI agents for customer service.

FAQ

How many stages can a Front sequence have? Between 1 and 10 stages (labeled A through J). Each stage after the first is a follow-up that sends only if the recipient hasn't replied within the number of days you set — for example, "Follow up in 7 days if no reply."

Does a Front sequence stop when someone replies? Yes. If a recipient replies to any stage, the subsequent stages won't send to their address. Automatic out-of-office replies are the exception and don't stop the sequence.

Can I personalize a Front sequence? Yes, with mail merge. Use {{variable}} syntax to pull CSV columns ({{sequence...}}) and contact custom fields ({{recipient...}}), with a fallback like {{variable|'there'}}. A sequence supports up to 200 recipients and up to 20 custom columns. An AI agent layer goes further by writing each step from live context rather than filling a blank.

What happened to Front AI Answers? Front's older AI Answers chatbot is now a legacy feature, "no longer available for purchase." Front directs new buyers to Autopilot instead, so plan around Autopilot for AI-powered replies.

Can I add AI to Front sequences without replacing Front? Yes. An AI agent layer like Macha connects to Front through the Macha–Front connector and runs on top of your existing inboxes and sequences — it doesn't replace them. Your sequences keep sending and chasing; the agent personalizes each step, reads reply intent, and drafts the follow-up. Macha's credits are consumed per AI action, never per resolution.

Ready to turn static follow-ups into smart, intent-aware ones? Start a free trial of Macha and connect it to your Front in minutes.