Following up on my previous 4A post, let’s look at how the methodology might work in a business process like onboarding a new supplier, something one company I know well did dozens of times a month. AI affords real opportunities for value generation in a process like this, especially where the existing workflow is manual, inconsistent, and error-prone.

Using the 4A methodology, we start with Analysis. How can AI provide insights from the data and systems we already have? This might include identifying existing suppliers that provide similar goods or services, or finding records from past work with a supplier (including previous attempts to onboard them). What’s crucial at the Analysis stage is that the tasks given to AI are generally low-risk and data-centric, helping us identify and address underlying issues in the data and systems on which further levels of automation will rely. And of course, the people who do this work gain an organic sense of what the AI does well and where it needs improvement.

Next comes Assistance. How can AI take away some of the mundane work so people can focus on more rewarding and intellectually demanding tasks? This might include drafting highly repetitive onboarding documents or searching external datasets to help build out a supplier profile. Risk is often slightly higher but still limited. User trust continues to develop organically. And importantly, AI errors at this stage provide invaluable feedback data.

Next is Augmentation – helping people do things they couldn’t do before. Those might include creating more sophisticated risk analyses that account for broader supply chain risks and dependencies or flagging potentially misleading or fraudulent information. People are still firmly in the loop here, but they’re increasingly leaning on AI. And through their experience at earlier 4A stages, they’ve developed an intuitive understanding of when – and how much – to trust it.

Finally, we might move on to partial or full Autonomy. This isn’t a pipe dream – Walmart has already deployed AI to negotiate directly with suppliers on its behalf, at least for relatively standardized agreements. For onboarding, standardized, low-spend suppliers could be handled end to end, with escalation rules for anything unusual. But getting to that level of capability requires building confidence in the technology, understanding its limitations, and establishing appropriate controls.

And that autonomous solution isn’t going to be appropriate for every supplier. When you’re onboarding a strategically important supplier or partner, you’re going to want your best people working on it. The good news is that with 4A, your team doesn’t have to choose between full autonomy and no AI at all. Instead, they can choose from a spectrum of AI capabilities – Analysis, Assistance, and Augmentation – to help them do their jobs better.

The goal of 4A isn’t maximal automation. It’s implementing AI in a way that maximizes value from the human-AI partnership.