Over the past two years, we have had the same conversation with dozens of organisations — government departments, hospitals, universities, medium-sized businesses, and NGOs. The conversation always starts the same way: "We know AI is important. We need to do something with it. Where do we start?"

The honest answer is that most organisations do not need to start with AI at all. They need to start with automation. And the distinction matters.

The Automation vs AI Confusion

AI is a specific technology — machine learning models, large language models, computer vision, and related tools that can perform tasks that previously required human intelligence. Automation is broader: any system that executes a process without human intervention.

Many of the things organisations want to accomplish — replacing manual Excel reports with live dashboards, routing incoming emails to the right person, automating approval workflows, sending scheduled reminders — do not require AI at all. They require workflow automation, which is a much simpler and cheaper category of technology.

The most expensive mistake organisations make is buying an AI solution for a problem that a simpler automation tool would have solved just as well for a tenth of the cost.

The Practical Framework: Find Your Highest-Value Manual Process

The right starting point is always a specific manual process — not a technology. Here is the question we ask every organisation: "What is the most time-consuming thing your team does every week that involves taking information from one place and putting it somewhere else?"

The answers we get most often:

• Compiling attendance data from Excel files into a payroll report
• Copying data from an email or form into a software system
• Generating the same weekly or monthly report manually from multiple data sources
• Sending follow-up emails or reminders that could be triggered automatically
• Routing documents or requests to the right person for approval
• Extracting data from scanned documents or images

Every one of these is automatable with existing tools — and most of them do not require AI.

The Tools Available Today (And What They Are Good For)

n8n and Zapier are workflow automation platforms — they connect applications and move data between them based on triggers and conditions. If data appears in system A and needs to go to system B with a transformation or notification in between, n8n or Zapier can do it. No code required for most use cases.

Make.com (formerly Integromat) is similar — a visual workflow builder that connects over 1,500 apps. Particularly well-suited for more complex multi-step workflows.

ChatGPT and Claude — large language models — are genuinely useful for drafting documents, summarising information, answering questions based on provided context, and classifying text. They are not useful for tasks that require real-time data access, precise arithmetic, or structured data processing.

Custom GPTs can be trained on your organisation's specific documents and policies — useful for internal knowledge bases, HR policy questions, and department-specific assistance.

Microsoft Copilot is worth considering for organisations already using Microsoft 365 — it integrates with Word, Excel, Teams, and Outlook, and can summarise documents, draft emails, and generate reports from data.

A Realistic Starting Point

Based on the organisations we have worked with, the highest-value first automation projects are typically:

• Attendance data from biometric devices → automated payroll input (saves 2–4 days per month for most organisations)
• Monthly management reports → automated data collection and report generation (saves 1–2 days per month)
• Incoming email triage → automated routing to responsible teams (saves hours per day for large teams)
• Document data extraction → structured database entry (relevant for banks, insurance, and government registration offices)

Start with one. Implement it well. Measure the time saved. Then repeat.

What We Do in Practice

When an organisation engages us for AI and automation consultancy, we start with a free workflow assessment — typically a 2–3 hour session where we walk through the team's daily activities and identify every manual, repetitive task. We then produce a prioritised list of automation opportunities with estimated implementation effort and time/cost savings for each.

The organisations that get the most value from automation are not the ones who spend the most — they are the ones who start with the right problems.