Most users are applying AI only on the surface. The potential is much bigger than that, but it requires us to finish homework first. When we automate a poor process, we scale rubbish. That’s just like selling products at a net loss. Growing the number of products sold will increase our top line but also worsen our bottom line. Automating poor manual workflows will pile up problems. Enhancing automation with AI is creating the perfect storm in no time.
Most of us have used AI in multiple applications without recognizing it. Search engines are one explicit use case everyone is familiar with. The providers are so kind to flag the summary on top as an AI feature, easy to recognize. This probably is more about marketing than transparency, but that’s not of importance to us right now. Few of the regular GPT users are paying customers, which is unfortunate. The difference between free software and paid subscriptions is often quality and depth, not just customer service. Diving deeper into the feature landscape of AI reveals many benefits, opportunities and in parallel overwhelming complexity. We need to be willing to invest into mastering this complexity to leave the AI-playground and turn it into a professional power tool. This complexity is the place where value creation starts.
“If all you have is a hammer, everything looks like a nail.” AI is a tool too, just like a big hammer, and, in this matter, it’s not any different from a hammer. From a different angle: A carpenter will always use wood to build; a locksmith will use steal. AI engineers will always try to maximize AI consumption. Hence, it is very important to approach this powerful technology with care and step by step, which is different from moving slowly. Automation follows a simple logic, always. First, try; then, learn, improve, standardize; and only after this iterative process start automating. Automation is great for standardized, well documented processes of repetitive tasks that we want to scale. Automation should provide a healthy balance between upfront investment and improved margins through better unit economics after the ramp up phase. AI can help us to automate more complex and more specific use cases and workflows that require vast data, for example.
AI demands data, access, and clear solution cases. Without the homework, also referred to as the initial investment, we won’t enjoy our shiny AI projects and sink all budget within weeks. AI is like ERP, CRM, etc. systems: the data quality defines most of the value added. Unfortunately, we are required to create structured data pools, lakes, etc. to become capable of leveraging AI effectively. The sensitivity of automated workflows for data quality is striking and pressures cultural awareness of all stakeholders. But, if we accept the challenge, we’ll be able to build powerful processes that enable our team to work faster and create better results, because we can cut much unwanted, manual work out of our busy, daily schedules.