AI Ready Model – Proof of Concept

Once you have been able to increase the knowledge in your team/organisation on AI tech and its protential, it is time to move on with a few identified potential use cases. As a result of the Labs phase you should have a list of potential use cases suitable for an implementation of AI tech within your organisation. 

From this list you should choose one or a few use cases that are candidates for a ”Proof of Concept” (PoC). A PoC means that you try out the technology and assumptions you have made in order to verify if it is possible to implement AI tech to get the results you desire in real life use cases.

A Proof of Concept is in all its essence a test and should be treated that way. However, if the PoC proves successful you should be ready to turn it into a production environment, so keep that in mind when planning and assigning resources. When prioritizing between different use cases to use for the PoC, you should look for cases that makes the most impact for the organisation or has a chance to prove real value. 

This way the PoC can potentially be used as a good example for the further initiative. Before starting work on the PoC you should define success criteria and what you want to evaluate with the PoC. These criteria should not only include technical aspects, but also implementation methodology, training, security aspects and organisational adoption. 

If possible the success criteria should also include KPIs for the business effects you wish to achieve. These could for instance be shorter time for customer enquiries, number of customer enquiries through-put during a given period of time, increased coding efficiency or whichever business gain you are looking to achieve with your AI initiative. 

Might also be that the initiative targets completely new business ideas or functions that the organisation wants to add or support through AI.

The whole intention of the PoC is to answer to the question of whether it is possible to implement AI technology with the choosen models and methodology. This is also why each PoC case needs to be evaluated against the success criteria defined. 

If the PoC for some reason is not successful, this is also of great value as long as we can learn from it and adjust potential errors made or challenges underestimated for the next case. This way the organisation will learn over time with real life cases in an agile manner. 

When a PoC implementation is considered successful, you will have a ”recipe” for how an AI use case can be implemented. A successful PoC implementation can therefore be used as a good example and should be lifted as such to give your initiative a good push and prove the potential value for the rest of the organisation using AI tech.

For a successful PoC implementation, you should be prepared to move it into production, so that you can make use of the resources spent and let the organisation draw the benefits from the implemented use case. This next phase will be covered in coming blog posts of this AI Ready series introducing the ”Redpill Linpro AI Ready Model”.