AI Automation with n8n

The skill clients pay for is spotting it, not building it.

• Mapping a process before touching a tool hands on
  Trigger, steps, decisions, exceptions. Most failed automations were built from an incomplete map.
• What should never be automated
  Judgement calls, apologies, anything where being wrong costs more than the saving.
• Costing it honestly
  Hours saved per month versus build hours and running cost — the calculation that wins the client.
• Where Pakistani businesses lose time
  Order intake on WhatsApp, invoice chasing, quotation writing, report compiling. Unglamorous and everywhere.

Nodes, data and the shape of a workflow.

• Self-hosting versus cloud hands on
  Installing n8n on a VPS, and the trade-off in cost, control and data residency.
• Triggers, actions and the data that flows between hands on
  Reading the item structure — where most beginner confusion lives.
• Branching, merging and loops hands on
  IF, Switch and batching, built so the logic is readable next month.
• Expressions and light JavaScript hands on
  Just enough code to reshape data between nodes.

APIs, sheets, mail, WhatsApp and databases.

• REST APIs and authentication hands on
  Keys, OAuth, headers, and reading someone else's documentation.
• Google Sheets, Drive and Gmail hands on
  The systems most small businesses actually run on.
• WhatsApp Business API hands on
  What is possible, what needs approval, and the rules on unsolicited messages.
• Databases and webhooks hands on
  Reading and writing directly, and receiving events from other systems.

Where a model earns its cost in a workflow.

• Calling a model from a workflow hands on
  Prompt, response, cost per run, and handling a refusal.
• Classify, extract, summarise, draft hands on
  The four jobs that account for nearly all business AI use.
• Reading documents hands on
  Invoices, CVs and forms into structured fields — the highest-value use for most clients.
• Keeping a human in the loop hands on
  Approval steps, and deciding what must never send itself.

The difference between a demo and something a business relies on.

• Idempotency and the double-run hands on
  Why a retried workflow must not email a customer twice or create two invoices.
• Error handling and alerting hands on
  Knowing it broke before the client does.
• Rate limits, quotas and backoff hands on
  Designing for the API that says no.
• Credentials and least privilege
  Handling a client's keys responsibly, and what to do the day one leaks.

Handover, documentation and getting paid.

• Documentation the client can follow hands on
  What it does, what it costs to run, what to do when it stops.
• Handover versus retainer
  Two honest business models, and being clear which one you are selling.
• Testing before it touches a customer hands on
  Dry runs, sandbox data, and the staged rollout.
• Knowing when to hand it back
  Automations that need constant babysitting are usually the wrong automation.

Before you start

• Free: APIs for Automation
  Strongly recommended before week one.
• Comfort reading technical documentation
  No programming required, but you will read a lot of API docs.

Where it leads

• Python Programming
  For when a workflow tool stops being enough and you need real code.
• Artificial Intelligence
  Deeper on the AI side of the same work.

Roles this prepares you for

• Automation Specialist
  In-house or agency.
• Freelance Automation Consultant
  Requires the business skills taught in the Freelancing course as well.
• Operations Analyst
  Improving how a business runs, with automation as the tool.

Interested in this programme?

Try the subject free first — there is no reason to pay before you know you like it.