Skip to main content
Business AI
How to Automate Tasks With AI
A non-technical guide to building AI workflows — no code required. Eight real automations you can set up this week using Zapier, Make, or n8n.
The opportunity
60%
Of jobs contain at least 30% of tasks that can be automated with current AI tools — per McKinsey Global Institute [McKinsey]
4hrs
Average weekly hours recovered per knowledge worker from automation of repetitive tasks with AI workflows [Salesforce]
$0
Code required to build a basic AI automation using Zapier, Make.com, or similar no-code tools — ideal for any small business getting started.
Traditional automation (like Zapier) is good at moving data between apps: "when someone fills a form, create a CRM record." AI automation goes one step further: it can understand, summarise, classify, and generate content as part of that workflow. At the far end, you move from scripted workflows into full-blown AI agents. "When a support email arrives, use AI to classify its urgency, draft a response, and route to the right team" is a single automated workflow.
What to automate first
Not everything should be automated. Use this framework to prioritise the right candidates — it pairs well with how to use AI at work, which covers the manual use cases worth automating later.
1
High frequency, low variation
Tasks you do the same way 10+ times per week. Responding to the same types of enquiries, formatting reports in the same template, sending the same onboarding emails. The more repetitive, the better the automation candidate.
2
Data transformation tasks
Moving data between systems, reformatting it, summarising it. Meeting transcripts to action items. Customer feedback to categorised insights. Survey responses to structured reports. AI excels at these conversions.
3
Clear start and end points
Automations work best when there's a definitive trigger ("form submitted", "email received", "new row in spreadsheet") and a clear output ("send email", "create task", "update record"). Fuzzy inputs and outputs break automations.
4
Low stakes errors acceptable
Start with automations where an occasional error is recoverable. Don't automate financial approvals or patient records on day one. Automate marketing emails, internal notifications, and data summaries first. Build trust before giving AI automation access to sensitive systems.
8 automations to build today
Triage and respond to support emails
Easy
TriggerNew email to support inbox
AI actionClassify urgency, categorise type, draft response using your FAQ docs, send if confidence >90% or escalate to human
Convert meeting recordings to action items
Easy
TriggerNew recording uploaded to Google Drive or Zoom
AI actionTranscribe with Whisper, extract decisions and action items, create tasks in Notion/Asana/Linear, email summary to attendees
Summarise news mentions and competitor updates
Easy
TriggerDaily at 7am
AI actionPull from Google Alerts RSS feeds, summarise each mention with AI, compile into one daily digest email to your inbox
CRM enrichment from inbound leads
Medium
TriggerNew lead submits contact form
AI actionResearch company from website URL, classify lead quality, write personalised follow-up email draft, create enriched CRM record
Social content from blog posts
Easy
TriggerNew blog post published to WordPress/Ghost
AI actionExtract key points, generate 5 social captions (LinkedIn, Twitter, Instagram), save to content calendar in Notion for review
Invoice and receipt processing
Medium
TriggerInvoice attached to email or uploaded to Drive
AI actionExtract vendor, amount, date, category using AI vision, create expense record in Xero/QuickBooks, alert if unusual amount
Product review monitoring and response
Easy
TriggerNew review on Google Business or Trustpilot
AI actionClassify sentiment, draft personalised response (different tone for positive vs negative), notify team for negative reviews before sending
Weekly reporting
Medium
TriggerEvery Friday at 4pm
AI actionPull data from CRM, analytics, and support tools, generate a narrative summary of key metrics with trend analysis, send to team Slack
Which tool to use
Zapier
Best for: Non-technical users
6,000+ app integrations, easiest interface, AI actions built-in. Limited flexibility for complex logic. Most expensive at scale.
From $20/mo (2,000 tasks)
Make (Integromat)
Best for: Complex workflows
Visual drag-and-drop builder with more powerful conditional logic. Better for multi-step automations. Better price-to-capability ratio than Zapier.
From $9/mo (10,000 ops)
n8n
Best for: Developers / open-source
Self-hostable, open-source, highly extensible. Best flexibility and lowest cost at scale. Requires technical setup. Preferred by developers.
Free (self-hosted) / $20/mo (cloud)
Where to start
Start with Zapier if you've never built an automation before — the interface is intuitive and their AI actions are pre-built. Once you're comfortable with the concept of triggers and actions, Move to Make for more complex workflows. Graduate to n8n if you have developer resources and need maximum flexibility or want to avoid per-task pricing.
FAQ
How long does it take to build an AI automation?
A basic automation (email triage, meeting summary) takes 1-2 hours to set up in Zapier or Make for a first-time user. A complex multi-step workflow (full CRM enrichment pipeline) takes a day or two. Most people see their first working automation in under an hour — the learning curve is front-loaded.
Do I need coding skills to automate with AI?
No, not for Zapier or Make. You configure triggers, actions, and AI prompts through a visual interface. Knowing Python or JavaScript opens up more flexibility (especially with n8n), but the major no-code platforms handle most business automation use cases without any coding.
What if the automation makes a mistake?
Build in a review step for anything consequential. For customer-facing automations, have AI draft responses and route to a human review queue rather than sending directly until you've validated accuracy. Most serious automation teams use a "human in the loop" step for the first 100 iterations before enabling fully autonomous runs.
Sources
[McKinsey] McKinsey Global Institute — "Jobs Lost, Jobs Gained" automation analysis (2023)
[Salesforce] Salesforce State of Work Report (2024)
04 — Don't watch from the outside
Veltrix Collective
Stay ahead of
the curve
the curve
Weekly briefings on AI tools, adoption trends, and what actually matters for practitioners. No hype. Just signal. Join readers navigating the shift.
Weekly, every Tuesday · No spam · Privacy policy · Unsubscribe anytime
Written by Luke Madden, founder of Veltrix Collective. Data synthesis and analysis by Vel.