Intro to AI Automation, Part 2
What you’ll be able to do: After this course you can decide which steps of a job need AI and which need a plain rule or a person, have Zapier draft every customer reply with a real signature, and get an email (and an automatic retry) when an automation fails. Then you build an HR agent that answers only from the handbook, an inquiry agent that collects the details before a person quotes, and an approval gate so an n8n AI agent asks before it emails a client. Last, you turn scanned purchase orders into reviewed data, put an after-hours voice agent on a law firm's phone line and build a client portal where each client sees only their own work.
- Who it’s for
- Anyone who finished Intro to AI Automation or has built one simple workflow or AI agent: office managers, owners, operations staff and team leads. No coding.
- Time needed
- About 91 minutes of hands-on practice: 9 modules of 6 to 14 minutes each.
- Cost
- Free. No credit card, no trial, and no sign-up needed to start.
- Level
- Beginner to Intermediate
- Prerequisites
- None for the practice: a web browser on a desktop or laptop. Helpful first: Intro to AI Automation. To repeat the modules in your own tools: an n8n account, an Airia account, a Guardian account and a Softr account.
- Format
- Self-paced and online. Each module runs in a practice copy of the real tool that checks every step, then a checklist walks you through doing it for real.
- Tools you practice in
- Zapier, n8n, Airia, Mixdesk, Guardian, DigiParser, ElevenLabs Agents and Softr.
Why this course
Part 1 took you from zero to a first workflow, a first AI agent and a web form whose leads get answered. Most teams hit the same wall right after that. The first automation works on a good day, and then a supplier's server is busy and a run fails without anyone noticing. An agent answers a question it should have said "I don't know" to, a draft goes to a client before anyone read it, and the jobs that eat the most time (paper forms, after-hours calls, "where is my project?" emails) are still done by hand.
This course takes the next step one job at a time, each with the AI tool that does it best and its own made-up business. Step 1: decide where AI belongs in an automation (n8n, Make or Zapier). Step 2: Zapier drafts every patient reply for a dental office and Signatoro signs it. Step 3: failure alerts and automatic retries in n8n for a florist's morning price update. Step 4: an HR policy agent in Airia that answers only from the handbook. Step 5: a Mixdesk AI agent that collects event-inquiry details before sales quotes. Step 6: Guardian in front of an n8n AI agent, so client emails wait for a person's OK. Step 7: scanned purchase orders read into reviewed fields in DigiParser. Step 8: an after-hours intake voice agent for a law firm in ElevenLabs. Step 9: a client portal in Softr for a marketing agency, tested as a client before anyone sees it.
The payoff: automations you can leave running, because a failure emails you and retries itself; agents that answer routine questions from your own documents, collect the details first and ask before they send; and paperwork, late-night calls and status questions handled without retyping or waiting until morning. You stay in charge the whole way: you decide which steps get AI, test every agent on a question it should not answer, review every value the AI reads from paper and approve every email to a client. Each module shows only its own tool, in a practice copy that checks your work.
Who this course is for
Start here after Intro to AI Automation. Three modules are beginner level (where AI belongs, DigiParser and Softr). The other six are intermediate and assume you have seen one workflow or agent, which Part 1 covers. Do them in order, or jump to the job your team needs most.
You can already
- Finished Intro to AI Automation, or built one simple workflow or AI agent before
- Have a repeat job at work you would like to hand to AI
- Nothing else: no code and no accounts to start (the practice runs in your browser)
After this course you can
- Decide which steps of a job need AI, a plain rule or a person
- Have AI draft replies with a real signature, and get an alert and a retry when an automation fails
- Build AI agents that answer only from your documents and collect details before a person steps in
- Make an AI agent ask before it emails anyone outside the team
- Turn scanned documents and after-hours calls into checked information, and give clients a portal that shows only their own work
What you’ll build
- Module 1: You sort six jobs for Harbor Lane Bookkeeping (made up) into plain rules, AI steps and human decisions. Then you walk through one AI automation from top to bottom: a trigger, one small AI step that picks a label and drafts a reply, a rule that acts on the label, a person who approves, and an action that saves a draft.
- Module 3: An error workflow called Error alerts: an Error Trigger followed by a Gmail email that names the workflow, the step that broke, the error message and a link to the failed run.
- Module 4: You are Maya, HR manager at Northwind Partners, a made-up 60-person company. In Airia you add the Northwind Employee Handbook as a data source with a clear name and description, upload the file, and build an HR Policy Agent that answers only from it, in three sentences or fewer, and names the section it used.
- Module 5: You run sales for a made-up event venue. You create an AI agent, write a persona with five details to collect and a rule against quoting prices or promising dates, and test it with a one-line inquiry.
- Module 6: You help Maya Chen, who runs Harbor Line Bookkeeping, a made-up firm whose n8n agent drafts about 60 invoice reminders a month.
- Module 7: You run the order desk at a made-up foodservice distributor. You build a purchase order parser, add a requested ship date, and add a review stage so every document waits for a person.
- Module 9: You run client services at a made-up marketing agency. You describe a client portal to Softr's AI, then preview it as a client, fix what you find, and test again.
- Module 1 · Automation basics · Beginner · ~6 min
When to Add AI to an Automation (n8n, Make or Zapier)
You’ll be able to: Step 1, know where AI belongs: sort six jobs into rules, AI steps and human decisions, and see how n8n, Make and Zapier each add one small AI step.
- Module 2 · Zapier · Intermediate · ~12 min
Zapier: ChatGPT Drafts Every Reply, Signatoro Signs It
You’ll be able to: Step 2, replies that sign themselves: build a four-step Zap where ChatGPT drafts each patient reply, Signatoro adds the real signature and Gmail saves it as a draft for a person to send.
- Module 3 · n8n · Intermediate · ~10 min
Get Alerted When an Automation Fails, and Let It Retry by Itself, using n8n
You’ll be able to: Step 3, automations you can leave running: build an n8n error workflow that emails you when a run fails, turn on Retry On Fail and retry a failed run from the Executions tab.
- Module 4 · Airia · Intermediate · ~10 min
Answer Employee Policy Questions Instantly using Airia
You’ll be able to: Step 4, an agent that answers from your documents: upload the employee handbook to Airia, build an HR agent that answers only from it, check it says "I can't find that" when it should, then publish it to the team.
- Module 5 · Mixdesk · Intermediate · ~7 min
Mixdesk: An AI Agent That Collects Inquiry Details Before You Follow Up
You’ll be able to: Step 5, collect the details before a person replies: build a Mixdesk AI agent that asks for the date, guest count and budget without quoting a price, test it and make it the first responder.
- Module 6 · Guardian · Intermediate · ~14 min
Guardian: Make Your n8n AI Agent Ask Before It Emails a Customer
You’ll be able to: Step 6, approve before it sends: write a Guardian policy so an n8n AI agent emails the team on its own, waits for your OK on client emails and never deletes.
- Module 7 · DigiParser · Beginner · ~9 min
DigiParser: Turn Scanned Customer POs Into a Reviewed Sales Order File
You’ll be able to: Step 7, stop retyping paper: read scanned purchase orders into the same fields with DigiParser, catch a handwritten change in review and export a checked file.
- Module 8 · ElevenLabs Agents · Intermediate · ~9 min
After-Hours Intake Voice Agent for a Law Firm
You’ll be able to: Step 8, answer the phone after hours: build an ElevenLabs voice agent that takes a law firm's after-hours intake calls, says it is AI, never gives legal advice and hands opinions to an attorney.
- Module 9 · Softr · Beginner · ~14 min
Softr: Build a Client Portal Where Each Client Sees Only Their Own Work
You’ll be able to: Step 9, a client app you can trust: describe a client portal to Softr's AI, preview it as a client and fix what the test finds, including one client seeing another's projects.
Prerequisites
- For the practice: a web browser. Each module runs in a copy of the real tool’s screens, so you need no account and no sign-up to start module 1.
- Helpful first: Intro to AI Automation.
- To do it for real: an n8n account, an Airia account, a Guardian account and a Softr account. Each module says which one it uses.
Time to complete
About 1.5 hours of hands-on practice across 9 modules (276 steps in all). Modules take 6 to 14 minutes each, so you can do one in a sitting and come back for the next.
Take it with you
- Intro to AI Automation (Part 1) · Start from zero: ChatGPT, a first n8n workflow, an AI agent and a web form.
- AI Agents 101 · More AI agents, from what an agent is to agents with tools and approvals.
Questions about this course
What should I learn after an intro to AI automation?
Learn to make automations you can trust when you are not watching: an alert when a run fails, AI agents that answer only from your own documents, and an approval step before anything reaches a client. Then use AI on the jobs that still eat the most time, like paper forms, phone calls and client status questions. This course covers each of those in one hands-on module.
Do I need to finish Part 1 first?
It helps but it is not required. If you have built one simple workflow or AI agent before, you can start here. If you are brand new, Intro to AI Automation (Part 1) starts from zero with ChatGPT, n8n and Lovable.
Which AI tools does Part 2 use?
One tool per job: Zapier with ChatGPT and Signatoro for signed replies, n8n for failure alerts, Airia for an HR policy agent, Mixdesk for an inquiry agent, Guardian with n8n for approvals, DigiParser for scanned purchase orders, ElevenLabs for a phone agent and Softr for a client portal. The first module compares n8n, Make and Zapier. Each module shows only its own tool.
How do I stop an AI agent from sending something wrong?
Give it rules and a person. In this course the HR agent answers only from the handbook and says when it cannot find an answer, the inquiry agent collects details but never quotes a price, and Guardian holds every email to a client until someone approves it. Test each agent on a question it should not answer before anyone else uses it.
Do I need paid accounts or coding skills?
No coding, and no accounts to start: every module is a practice copy in your browser that checks your work. To do it for real, each module lists what you need. Most of the tools have free plans or trials, and plans change, so check each site.
How long does the course take?
The course takes about 91 minutes of hands-on practice across 9 modules of 6 to 14 minutes each. It is self-paced: do one module a day, or start with the job that slows your team down most.
Want help doing this for real? Book a 1:1 session with Robert ($85/hour) →