- Most AI pilots stall because nobody owns turning them into daily workflows, not because the tools are weak.
- Start with one costly, repetitive process, measure its baseline, then automate it end to end.
- Put a named person in charge of implementation, and report hours and cost saved every month.
- Scale only after the first workflow shows savings on the profit and loss account.
Almost every company in Hong Kong now has staff using ChatGPT, Copilot or a similar assistant. Yet when finance asks what AI has actually saved, the answer is usually vague. The gap between "we use AI" and "AI cut our costs" is an implementation gap, and it can be closed with a disciplined process.
Step 1: pick one process that costs real money
Look for work that is repetitive, rules-based and done by several people: invoice entry, customer enquiries, report preparation, sales follow-up. Estimate the hours it takes each month and multiply by loaded staff cost. That number is your baseline and your business case.
Avoid starting with the most impressive idea. Start with the most measurable one.
Step 2: map the workflow before choosing tools
Write down every step, every hand-off and every system the work touches. Most savings come from removing steps, not from adding a chatbot to a broken process. Once the workflow is clear, the tool choice is usually obvious: an enterprise AI assistant, a workflow automation platform, or a connection between your existing systems.
Step 3: give implementation an owner
Pilots fail when AI is everyone's side project. Appoint an implementation lead, internal or external, whose job is to build the automation, train the team and fix what breaks. For larger programmes, an implementation lead plus one or two automation specialists can work through several workflows over six to twelve months.
Step 4: measure, report, then scale
Track hours saved, error rates and turnaround time against the baseline from step 1, and report them monthly by department. When the first workflow shows a clear saving, use the same team and method on the next one. This is how AI moves from an experiment to a line on the profit and loss account.
Where GI Executive helps
We provide AI Implementation Teams and individual AI talent on an outsourced, project-based or permanent basis, so you can add implementation capacity without opening a new headcount.
Want this working in your company?
Tell us how your team works today. Within two business days you get a recommended hiring model and a scoped proposal.
Talk to us