Step by step
Define which email types you get and how they should be sorted/prioritised (categories, urgency, standard replies).
Before automating, define which mail types you actually receive and how they should be sorted. This structure is the basis – without it the automation filters aimlessly.
Example promptI want to automate my email inbox with AI. Help me define a system: which categories of email do I typically receive (e.g. enquiries, invoices, newsletters, support)? Suggest 5–7 sensible labels, prioritisation rules and for which types automatic reply drafts make sense. My context: [job/industry].
Write reusable reply templates in your personal tone for the most common email types.
Reusable reply templates in your tone of voice save the most time. Phrase them for the most common cases so the AI can build on them.
Example promptCreate reusable email reply templates in my tone ([friendly-professional/casual/formal]) for these common cases: [e.g. appointment request, request a quote, decline, follow-up question]. Each template with placeholders for individual details, polite and concise.
Build an automation that reads new emails, categorises them with AI, applies the right label and creates a reply draft.
When building the automation, the rule is: first a simple flow (read, categorize, draft), then test. An overloaded first attempt is hard to understand.
Example promptWalk me through building an automation in Zapier step by step: the trigger is a new email in [Gmail/Outlook]. An AI step should assign the email to one of my categories [categories], set the matching label and, for the intended types, create a reply draft from my template. Which steps, apps and fields do I need?
Optional: map more complex flows (e.g. save attachments, create tasks, forwarding) visually in Make.
More complex flows like saving attachments or creating tasks are only worth it once the basics run reliably. Build them in step by step.
Example promptI want to extend my email workflow in Make: for emails of category [e.g. invoice] the attachment should be saved automatically to [cloud folder] and a task created in [tool]; for [category] a forward to [person]. Describe the scenario with the needed modules and filters.
Tips & common mistakes
- Most common mistake: automating too much at once. Start with a simple flow and expand it bit by bit.
- Let the AI create drafts instead of sending directly – that keeps you in control.
- Treat confidential mail with caution and check the data protection of the tools you use.
- Define clear categories first. Without clean sorting rules the automation works aimlessly.
Frequently asked questions
Is this doable without tech knowledge?
Yes. Tools like Zapier or Make work via building blocks, and the AI describes the setup step by step. Start with a simple flow.
Does the AI send mail automatically?
Usually it only creates drafts that you check before sending. That way you keep control over what goes out.
What about data protection?
Handle confidential content sensitively and check how the providers deal with your data. For sensitive mail, caution is warranted.
Does this work without technical skills?
Yes. Tools like Zapier or Make work through building blocks, and the AI describes the setup step by step. Start with a simple flow.
Does the AI send mail on its own?
As a rule it only creates drafts that you check before sending. That way you keep control over what goes out.
What about data protection?
Handle confidential content sensitively and check how the providers treat your data. With sensitive mail, caution is warranted.
Where do I best start?
With a single clear rule – for example: automatically sort newsletters into a label. Once that runs, you expand step by step.