Skip to content
Replicer
No-code AI workflows

AI automations without a line of code

A visual builder to decide what happens when a call ends, a message arrives or a set time comes round. Messages, AI steps, conditions and waits snap together as blocks, and you test them before switching on.

Confirm after the call

Active
  1. Trigger

    Call finished

    Outcome: answered · Reception agent

  2. Step 1

    Extract data with AI

    Name, treatment, preferred date

  3. Step 2

    Appointment booked?

    Yes: confirm · No: alert the team

  4. Step 3

    Confirm on WhatsApp

    Approved template, practice name

  5. Step 4

    Alert the front desk

    Email with summary and transcript

  • Triggers on real events

    Calls, WhatsApp and email messages, incoming webhooks and schedules.

  • Built-in AI steps

    Generate text, extract data and classify requests inside the flow.

  • Waits of up to 90 days

    For reminders, periodic recalls and check-ins weeks later.

  • Versions and history

    Test changes, publish them and review every run.

Triggers

The flow starts when something happens

A call that has ended or been analysed, filtered by agent, direction and outcome. A WhatsApp message or email, optionally only with a keyword or on first contact. A webhook from your website, a schedule or a manual start.

  • Call ended or analysed, filtered by agent, direction and outcome
  • Outcomes: answered, no answer, voicemail, failed
  • WhatsApp or email: any message, a keyword or a first message
  • Incoming webhook, hourly, daily or weekly schedule, or manual start

When the automation starts

  • Call finished or analysed

    Filter by agent, direction and outcome

  • WhatsApp or email received

    Any message, a keyword or the first message

  • Inbound webhook

    From forms, business software and other apps

  • Schedule

    Hourly, daily or weekly

  • Manual start

    Whenever you choose, for tests or edge cases

Actions

Messages, AI and conditions as building blocks

Every step is a block: WhatsApp with a template or free text, SMS, email, team alerts and AI steps that generate, extract or classify. Conditions and waits decide what happens and when. If the 24-hour WhatsApp window has closed, a dedicated branch takes over.

  • WhatsApp by template or free text, with a branch for a closed window
  • SMS from an alphanumeric sender and email from your domain
  • AI steps: generate, extract, classify
  • Conditions, waits of up to 90 days, team alerts and an end step

What the automation can do

  • Send a WhatsApp message

    Approved template or free text within 24 hours

  • Send an SMS

    With your business name as the sender

  • Send an email or alert the team

    From your domain, to contacts or chosen colleagues

  • Generate, extract or classify with AI

    Summaries, structured data, categories

  • Conditions and waits

    Different branches and pauses of up to 90 days

Control

Edit, test and switch on safely

You work on a draft while the live automation keeps running its published version. Test runs use the draft and simulate sending, so nobody receives anything. The run history shows every step, and the number of active automations depends on your plan.

  • Published versions, with the option to restore an earlier one
  • Test runs on the draft, with simulated sending
  • Run history with the outcome of every step
  • Active automations according to your plan

Run history

Version 7
  • Confirm after the call

    Completed · 5 steps · 2.1 s

  • Confirm after the call

    Completed · branch: no answer

  • Test run

    Sample data · no messages sent

  • Version 7 published

    Added a 2-day wait before the callback

AI business process automation, no code required

A few examples of flows you can build with the available blocks, starting from events that already happen every day in your business.

  • Unanswered call

    If a campaign can't reach a contact, an SMS under your name goes out inviting them to call the office back.

  • Periodic recall

    After an appointment, a 90-day wait and then a WhatsApp template reminding the customer to book their next check-up.

  • Request triage

    An incoming email is classified by AI: complaints trigger an immediate team alert with a summary and the customer's details.

  • New lead from your website

    Your website form sends the data to a webhook: the contact gets a WhatsApp template and sales gets an alert with the extracted details.

  • Recap after a consultation

    When the call ends, AI extracts the key points and the customer gets an email listing the documents to prepare.

  • First WhatsApp message

    When a new contact writes for the first time, AI generates a welcome reply and your team gets a heads-up.

FAQ

Frequently asked questions

What are Replicer's AI automations?

They're flows that start from an event, such as a call ending or a message arriving, and run a series of steps: sending WhatsApp, SMS or email, alerting your team, or using AI to generate text, extract data or classify requests. You build them in a visual editor, without code.

Can an automation call a customer or book an appointment?

No. Automations handle messages, AI steps, conditions, waits and alerts. Calls and bookings are the voice agent's job: during the conversation it checks availability, books and can update connected apps such as HubSpot. The automation takes over afterwards, for example with confirmations and reminders.

Can I connect automations to my other tools?

Yes, inbound: a webhook can start a flow from your website, a contact form or your business software. To write data into other tools, use the agent's in-call tools, such as webhooks and apps connected through Composio, or Replicer's REST API.

How many automations can I switch on?

It depends on your plan: each plan includes a number of active automations. You can prepare and test flows, then switch on the ones you need within your plan's limit. Details are on the pricing page.

How do I test an automation before switching it on?

A test run starts the draft with sample data, such as a recent call or a message, and simulates sending, so nobody receives anything. Each publish creates a new version, and the run history shows which steps ran for every start and with what result.

Build your first AI automation

Start from an event you know well, such as a finished call, and add steps one at a time. Test it before switching it on.

Guided setup · Human support when you need it