You Want To Use AI. Now What?
Artificial intelligence and cognitive automation are already top of mind for most business professionals. Boards and C-suites are increasingly discussing corporate AI strategy. Executives across organizations are excited about AIs potential and are keen to use it.
In our work with clients across industries around the globe, we have faced this question multiple times. Identifying the right use case for an AI or cognitive technology stumps people. We are frequently asked, how can an enterprise use a particular technology. The short answer is “We don’t know”. We don’t know the business as well as our clients do. We don’t know what are the real pain points. Sure, there is a list of use cases solved for similar companies, but that is just the tip of the iceberg.
Figuring out the fit of best use cases ready for automation, with the best technology to automate is a three stage process. The crux is to start with the use case, rather than technology.
We have written before about our value of “Earning customers’ trust before their revenue.” Once we have a customers’ trust, we engage with them in identifying the best use cases for cognitive automation and other AI technologies. While it is different for each situation, here is the broad process we follow.
1. Start with Bottomline
Learning about a new technology or a new, powerful idea, is intoxicating. It is natural to be excited about the potential uses of this technology and to figure out how it can be applied to your world. Some times over eager sales personnel push this idea hard. However, this technology may not be a good fit with you at all.
Let’s start with a simple assumption – everything is automatable. We will question this assumption later, but its a great starting point.
Then a simple question follows. What workflow – a process, decision, or task – if automated, will have a transformational impact on bottomline, or the success metric for your team? This impact could show up as saved costs, increased revenue or simply a different way of doing business that makes your team smarter or more nimble. This analysis must include not only the existing workflows, but also the workflows that are currently thought to be prohibitively expensive, non-scalable or too time-taking to be useful.
It takes a certain creativity to do this but its crucial to start with the bottomline. These projects easily get budgets, organizational buy-in, rapid deployment and admiration. Starting with bottomline is a win-win situation for your team’s efforts, your organization and the technology provider you will end up working with. The added advantage is that stakeholders get involved from day one.
2. Classify the Problem
Lets say there is a prioritized list of most transformational workflows ready for automation. The following, simple decision tree helps in classifying the workflows:
- Is there a simple set of computer rules for this workflow? For example, consider the task of collecting prices of Duracell batteries from 70 different ecommerce platforms. These workflows are ripe for Robotic Process Automation.
- Does the workflow depend on large amount of structured data? For example, look up the now-famous use case of show recommendations on Netflix. If there is sufficient data, and the data is structured, then the workflow could really benefit from traditional AI algorithms like Deep Learning.
- Can an entry level professional or an intern complete the workflow, at least partly? For example, think about the workflow of reading through 1,000s of documents to find answers to specific questions. Workflows or parts of workflows that do not need human creativity or judgment, no matter how complex they are, can use Cognitive Automation.
This is oversimplified, of course. However, this kind of classification leads to a comprehensive strategy for enterprise automation. A lot of workflows will not belong to any of the above categories, but there will be many that are ready to be automated at different levels of sophistication.
3. Now, Select Vendor(s)
For each use case now, it should be easy to research and invite technology providers for further conversations. In the AI or automation space, there are a lot of companies that focus on a sector, on a particular application or on an application in a sector. If your use cases have a fit like this, such vendors could be a good starting point.
In most cases, your use cases will be unique to you. Even otherwise, you want to meet with general purpose providers. For instance, cognitive automation companies like IBM Watson and Coseer provide bespoke solutions that configure and train for each use case and client.
It may so happen that the vendor you loved did not make this cut. Its still better for them to hear a polite no, than to waste time trying to figure out the right use case, or go and try to solve a problem nobody cares about.
If you want to discuss how we can help you with your automation links please setup a meeting here. If you want to explore this on your own, this next generation knowledge management framework may be a good starting point to get your creative juices flowing.