The AI Product Management: An Evolutionary role of a traditional Product Manager

Why can’t Data Scientists build a whole AI product? Anyway, this is the job for they are being hired…

When I talk about AI in an organization setting, it usually means making some AI based application for our users. Most of the companies would advertise for a the Data Scientist’s job having expert level skills of data analysis, Machine learning, Software engineering, and data engineering to make a ship-able product end to end. So, what is the role of a Product managers in ML or AI. The answer is very simple. If we take the example of a mobile based start up, then a mobile developer can build up the application with the concept of the app that he can define or the CEO of the company can define, but when it comes to a more established and matured organisation, then the Product manager used to manage the product while defining Vision and Leadership, Product Life Cycle Management, Product Strategy and Market Research, Business Model and Financials, Product Roadmap and User Experience and Product Backlog. As organizations have access to massive amounts of data, AI enables them to draw insights from the data at a much larger (as well as more granular) scale than before. It could be customizing an individual user’s experience or to create new product that doesn’t currently exist.

What the Industry demands from a traditional Product Manager in all these respective key areas to become a Next Generation Product Manager, I mean to say an Artificial Intelligence Product Manager, as the market has been occupied by the Artificial Intelligence and is drastically moving towards the Augmented Analytics.

Product Managers are the Visionaries

A PM knows the market need, how the product should be conceptualized to solve the daily problem met by the people or the end users, which technology is the best to solve that problem, guiding the product to build and product-ionize the product in the proper environment- are some of the keep responsibilities of a PM. Here, the PM should not accept the AI blindly as it is trending in the market, but to access the requirements really seeking AI implementation, because making a AI product success is not so easy and not a job of one iteration or single time implementation like the traditional web or mobile applications. We all agree that appropriate use of ML algorithms is essential to differentiate the product and delivering a better value proposition to its customers. It’s a separate function that requires business acumen, a deep understanding of UX design, knowledge of ML algorithms, software engineering and off course the product knowledge. Product managers are responsible for setting a product vision, defining a product strategy and developing a roadmap that meets both company goals and user needs.

PM’s focus more on the problem and less on the solution. PM is the one who would define WHY are we building this product (product vision) but AI based PMs should have the technology acumen to understand that HOW. (product strategy and goals)

PM is the Product Concept Formulator, where a PM gets involved in below key areas of a Product Development:

Empathize: Behind every AI based product, there is a human centric emotion. A good product manager would like to build a product that will make the life easier either for themselves or their user. Thinking of a Dogbot didn’t originate from the motivation to build a great tool in NLP domain but it originated when someone empathize with the end-users and asked the question that help dog owners and dog lovers with dog’s health issues, dog adoption, dog products and NGOs.

Define:

AI based Product Management is not much different from building other software, but it requires some more things to learn. Its the duty of a PM to ask the right questions that ML can answer. Use ML and AI as a tool to answer those questions. How do we bring the best possible product to market and grow our business? How the result will be used, what your model should predict and how it should be calibrated, what data you collect and process, what algorithms you test and many other questions.

Being a PM, first i need to define what am I trying to solve here? how this new product will solve any problem for me or my users?

This is also the phase where you will define how the product success be measured? how is ROI defined? what is the metrics you are going to measure after each roll out of a product to define the success or otherwise?

PMs should focus on identifying the problem and it is entirely the responsibility of data scientist/ML engineers to identify the right ML algorithm based on the problem statement provided by PM.

Ideate: In this phase of the product management process, new suggestions, ideas and feature requests are captured as part of the product backlog, providing a good source of inspiration for product’s evolution, and the good ideas should be locked down and developed further. This is the ultimate responsibility of a AI product manager. Product manager will work with the data scientists and the data engineers to figure out about data collection, choice of ML algorithm, integrating APIs etc.

Prototype: Strategic Prototyping is used instead of simple prototyping in AI projects to decide which ML algorithm they need to use to answer the question and whether the picked algorithm is enough or further iterations are needed to declare the product as success.

Test: This is the combined work of data scientists and data engineers. We need to test if the chosen ML algorithm solves the problem? does it work with the production data? are there any glitches which need tone solved urgently? We should keep this in mind that no chose algorithm is the perfect one and it might need several iterations before the final product or as the updated versions of the same product later on.

This is the changing role where a AI based product managers usually has to get involved in the prototyping or testing phase while understanding the ML and engineering language to communicate effectively.

Productization: This is also the combined effort of data scientist and data engineers under the direction of product manager. We must reach to certain decisions about the characteristics of AI workloads and requirements for product ionizing AI model such as, data management, interchangeability, security, elasticity, and usability as well as keeping in mind the considerations for architecting AI pipelines for data generation (data prep and feature extraction), model training, and model serving. Product manager have to sit with data engineers, business heads and data scientists to make the tough decisions about tech stack and requirements to ship the best AI product.

The ability to scale AI to take on massive opportunities to engage and connect machine-to-machine or consumer-to- machine is the only method to show AI’s return on investment. So, a PM with the sound knowledge of ML and AI is required to make an investment in AI to bring value.