5 steps to avoiding wrong AI projects implementation

Share on facebook
Share on twitter
Share on linkedin
Share on pinterest
Share on whatsapp

The need to implement latest technologies and new approaches into different levels of the organisation structure and operational processes had never been seen so clearly before. It is the only way to stay in business and keep on growing. Numerous experts of the field never tire of reminding that the industry as well as service sectors should turn their attention to artificial intelligence.  And in particular that they should invest into the AI projects implementation. But why?

 

Benefits vs risks of AI projects

 

Let’s start with the term  “Artificial Intelligence (AI) projects”. This term brings under its umbrella a very large scope of processes and activities. In one way or another, all of them are meant to solve organizational and operational challenges. From communicating with clients to generating quotations, from safety observations at the factory to safety traffic counter, from benchmarking to project management. Anything that implies analyzing and combining different data points to come up with new information, could be improved by AI techniques. Lastly, AI-based solutions are also the key for data analysis and strategic planning. This analytics allow you to identify the problematic fields that keep your business down. Besides, they offer the smartest ways of improvement.

 

However, it is critical to keep in mind that digitalization and in particular digital transformation can pose risks to businesses if it is not implemented correctly. One of the most recent surveys (source) carried out among the top-management and senior executives of leading industry players around the globe shows raising concerns when talking about digital transformation of processes and operations. The latest Harvard Business Review study displays an unacceptable number of failed business transformation initiatives. Almost 70% of them did not reach their goals. Speaking in concrete numbers over $900 billion went to waste.

 

Practical steps for successful AI project implementation

“It is not only about the technology, but also 

about organization processes and people” 

 

How to make sure that your company avoids wrong implementing of AI system, that it will get useful insights for future business processes. How to guarantee that data science will bring to life your ideas by means of successful implementing of AI projects.

 

1. Focus on overall strategy to guide your AI efforts

 

Prior to any drastic changes driven by such a powerful tool as artificial intelligence, the business should analyze its strong and weak points. It should set clear goals and measure the available sources. It is fundamentally important to get that analytics and to know where you are now and where you want to go. The great piece of news is that digital data analysis can support you on that task. It will make it easier to value and frame the starting point.

Once it is done, turn to the actual business problems and focus on the specific business cases that have the data needed for efficient analysis and implementation. Do not be overwhelmed by the myriad of possibilities that AI brings to you. Take on specific problem in your business which has the following characteristics:

  • there is a clear question to answer. (How many rooms I will book in two weeks’ time in our hotels in Mallorca?)
  • you have the data needed to analyze the problem. (Historical pricing and booking data, weather forecast.)
  • you know how to measure the result. (Bookings done over a period of two weeks vs. bookings estimated two weeks before)

 

 

2. Build your data lake  incrementally

 

Conventional wisdom and multiple resources identify a simple idea: to get great results one should be well prepared. In other words, to get AI algorithms work we need to collect a large quantity of high-quality data. And that is true! But at AQUILES Solutions we strongly believe in the concept that “A journey of a thousand miles begins with a single step”.

To create an extensive high-quality database takes a lot of time. Taking that much time for simply collecting the information is unnecessary, inefficient and a costly decision. Instead, ensure you build the data lake incrementally. The benefits of doing so are faster results, clean data from the start and having the dataset organised, updated and expanded organically.

 

Example:

As a practical example from the healthcare industry, we can take the topic of capacity management of critical care beds. To improve the capacity management models and to forecast the workload of the critical care units for a given week, hospitals could introduce an AI-based system. The first step of designing this model and putting it in practice is to use the historical entry times data from the critical care units. Next step is to increase the data lake with the length of hospital stay data to estimate occupancy of the beds at ICU. Introducing new overall occupancy rates to the data lake on a regular basis will increase the prediction efficiency for bed planning. As a consequence, it will reduce the threat of intensive care nurses’ burnout and deliver effective and safe health care.

 

 

3. Involve your domain experts in the project

 

Any projects that involve data analysis, digitalisation of the processes, or digital transformation are always time-consuming and often unrealistic for internal human resources of the companies and industry. Outsourcing such projects allows to get highly specialised experts, attract new skills tailored to solving certain tasks and focus internal resources on the core activities of the company.

However, numerous businesses faced the situation when external experts were reducing the optimization solutions to a limited set of actions called best practices as to a panacea. Hopefully, one-size-fits-all solutions are the artifacts of the past. Outsourcing artificial intelligence projects does not mean to leave the whole process loose and in the full hands of Digital Transformation Consulting Companies. Involve your domain experts in the AI-based project to ensure two things:

  • The optimization and digital solutions are best tailored to the needs of the company
  • The knowledge and skills related to the tools and applied actions stay within the internal team. That will facilitate the basic maintenance and general understanding of the processes.

 

 

4. Refine your model as you move forward

 

Start simple and then scale it up as your needs and available resources are growing. AI projects may end up becoming super complex and overwhelming. Especially if we are talking about major intellectual models for which a large data base is required. Observing, collecting and analyzing data may take up to 6 months of work. Take that elephant and split it into smaller pieces.

As long as you have a clear objectives of what you want to have as a result, there is no need to build the entire model from scratch to a polished version. On the contrary, developing and refining your model on the move will allow you to make it more adjustable to the needs that pop-up in the process of implementing AI-project.

 

Example:

Let´s turn to another example when AI technologies services the real world and customer experience in particular. Say, you have only one external conversation email address to simplify the communication process for your clients and partners. And at some point you see the urgent need to implement an email classification system to optimize the communication flow.

First step of designing the tailored email classification model would be the classification based on the categories and departments where the email should be directed: operations, sales department or procurement. Next step of refining this model will be defining the sector within those departments. In other words, directing particular email to the team or team member who will tackle the topic. Let´s say: requests, claims or proposals. This algorithm allows any classification you find useful for your project.

 

 

5. Advanced flexibility of software model

 

Artificial Intelligence projects are implemented to support and foster business development. And in order to achieve that goal, the system that runs the AI-based solutions should not only be refined along the way but should also remain flexible and capable of extending alongside with the needs of the projects. AI projects succeed when they are open to the improvements suggested by the users who deal with the system daily.

Before you start the project, specify the software model that will be designed. Make sure that your provider is building an open system applicable to different purposes of your project.  Knowing who will use the system and how they will do it, is critical at this stage. It will ensure the usability and successful implementation of the AI project. And open model system will allow all team members to use it on a regular basis.

Such open systems may range from desktop apps to Matlab applications, AutoCAD, etc.

 

Example:

Does your team use Excel programme as a main tool to carry out their daily work? And you feel they can do it in a more intelligent way? You are right! Consider creating an Excel function that will run your model. Doing so, will allow your team to integrate it into their models and most likely use it for further purposes. This application flexibility of the existing model will ease a constant improvement of the processes. (For more ideas check our case study: PDF to Excel Converter)

 

Conclusion

 

In order to reach and/or maintain the leading positions in your market, the business must stay intelligent. As well as to move from problem-solving approach to proactive and self-driving mentality. The companies must search for ways of how they can use AI to get better market positioning.  On one hand, to meet a constantly increasing demand of the customers. On the other hand, to adapt to more sophisticated requirements of production line, waste control management and public safety. The implementation of the Artificial Intelligence projects with their classification, prediction, automation and optimization  solutions based on data analytics will reach out a helping hand to achieve all the market requests.

 

Follow the simple steps that we identify in this article. It will ensure that you are on track with your objectives that were set in the beginning of the AI project. It will make sure that the whole process of implementing AI is under control. And, lastly, it will assure that the tech solution that you apply now will bring more value to the future projects.

You might also like

AQUILES Solutions at Iberquimia Digital
This week AQUILES Solutions is participating in the second edition...
5 steps to avoiding wrong AI projects implementation
The need to implement latest technologies and new approaches into...
Industry 4.0 solutions
The custom service automation company will present its applications and...