MELDS – FIVE COMPONENTS TO HELP SUCCESSFUL AI IMPLEMENTATION

MELDS is an abbreviation for the five components: Mind, Experimentation, Leadership, Data, and Skills. These five components are invented for management purposes and should be considered when embarking on an AI journey. MELDS can be a valuable tool helping you reimagining business and become successful within AI and ML.

MARTIN BÖRJESSON | PARTNER, INDUSTRY LEAD: LOGISTIC & TRANSPORT | APRIL | 2019

MELDS – key components to start your AI journey

MELDS was invented and put together by Daugherty and Wilson in their book “Human + Machine” (Reimagining work in the age of AI). MELDS is a truly great way to address the required components, and also to discuss what kind of conditions need to be established and maintained to become successful with AI in any enterprise and organization.

MELDS components

1. Mindset

The sequence of MELDS is not random. Many organizations are actually missing an updated mindset, which is why they are not started on their AI-journey. An updated mindset is a compulsory component from executive management and, in the ideal world, down through the entire organization.

When updating the organizational mindset, the focus needs to be on an entire business process that is truly “reimagined”. This means that we are looking for results in a brand new way, scrutinizing our old business processes and going for new transformative results.

Once this step is covered, the organization is ready to move to the next phase: Experimentation.

2. Experimentation

For many organizations, the phase of experimentation can be an incredibly hard nut to crack, as this can become too close to playing around with their funds. According to the authors of MELDS and myself, this should be related to R&D instead.

All companies are doing such activities, searching for better products and services, and here AI can be the infuse they have been looking for but simply didn’t know what it was capable of. Even though experimentation should always be done carefully, it is a necessity, especially as it can be difficult to assess the results that will come out from the other end of an AI project.

Experiments almost never take their full scope of business into play, however, an organization that managed this was Coca Cola.

By taking a look at their vending machines, Coca Cola was able to make decisions on replenishments values, sales forecasts, planning of deliveries, etc, all calculated out of an image taken by a kiosk owner and uploaded through an app. That is cheap, simple and very effective experimentation, which could not have happened without excellent leadership.

3. Leadership

Leadership is perhaps the most difficult component to succeed with in any traditional organization today. Many organizations have not been pursuing emerging technologies and truly speaking, they are sometimes not very well updated.

Here, tech-savvy internal and external sources of inspiration play an important role. Tools and techniques to bring down the technology to a level where a leadership team feels both comfortable and enthusiastic about its potential, are today considered as an art.

External factors like risk, funds, sense of urgency, age, political and economic mandate – the list is long – will affect how to play the art every time.

This also means that in order to become successful leaders with AI, a solid investment in a true understanding of current business problems and pains is required, as well as connecting those directly with the pre-identified solutions with access to data.

4. Data

Data is the fuel and a new energy to be valued higher than many other assets in the years to come. Data can be images, video, and structured data tables, and does not necessarily have to be very deep and long to produce value – see for instance the Coca-Cola example mentioned earlier.

What is more important is for companies to establish data organizations and skills to bring out the new value, products, and services. This becomes possible due to the data available, however, it does not necessarily need to be internally produced data. It can also be public sources like weather forecasts, which was a key element in the Coca-Cola case. Here warm or cold weather could indicate to Coca Colas supply chain operations if consumers would be more thirsty or not.

Utilizing data will also bring out discussions that we need to have, as the ethics and privacy perspectives will automatically be alerted. Cambridge Analytica has proved that by only 70 likes, they could establish very precise profiles about gender, religion, political party preferences, etc., with a very high degree of accuracy.

This is also exactly why Trump, during his campaign, went 5 times to the smaller state Wisconsin. The data of Trumps analytical staff revealed that 60 – 70.000 Wisconsin voters were likely to move from Clinton to Trump. Due to extensive electioneering, Trump ended up winning the state with 22.000 votes – a big surprise to many, but not to the people running Trump’s campaign – they had it in their data and they had the skills to extract value from the data.

5. Skills

Skills are the fifth element and for now, the one causing the least problems. However, as the development and implementation of AI increases and become more common, problems around skills will increase as well.

The development and increasing implementation of AI will change the workload and workforce composition in the organizations. This means that organizations have to, and some already had, remove the human to human interaction – for instance in retail. Organizations are moving to e-commerce and the skills required here will rely on how to personalize the messages we meet on the web. For this, we will need people who can analyze language, mood, temper and general human conditions in order to bring the correct response from AI to the consumer.

These competencies are not yet recognized today but eventually, as we include AI into all our different business processes, it will become clear that the skills required, are not the skills we used to have. Now we have to deliver a new set of skills via AI or through its experiences.

Summary

MELDS is a great concept and very easy to remember. It has a stickiness which should make a lot of sense to many executives and stakeholders when starting to work with AI and ML. Here MELDS can be a useful tool, helping organizations turning the promise of AI into real business value. Let’s all remember this, once we hit the AI road for success!

You might also like…