Prasana Murthi is a Head of Product at Receipt Bank, who are industry leading providers of software accounting solutions for SME sized enterprises. John, our Specialist Product Management Recruitment Consultant at Consortia managed to catch up with Prasana to discuss some of the complexities involved in Product Management with a particular focus on how Artificial Intelligence is a now a key focus when building World class Products in today’s technology-driven marketplace. Considering AI is moving such a rapidly growing trend, we posed this question – How will AI move Product Management forward in the years to come?
About Prasana Murthi
The initial discussion focused on Prasana's background and experience within the Product sector and, more specifically, artificial intelligence (AI).
With a strong engineering background, Prasana graduated in the Year 2000, during the dot-com crash. His first job was in Telecoms for Alcatel-Lucent, who is now owned by Nokia. Although this was a software engineering role, over time, it gradually migrated towards the Product Management side. His next role at IBM exposed him to IBM Watson which is IBM's AI/Machine learning offering. It was during his role with IBM where he delved further into the study of how AI products can be marketed whilst also studying for an MBA from the leading CASS Business School. During the course of his conversation with John, Consortia’s Product Management Recruitment Consultant: Prasana explained that it was "during this process, I understood the technology, I understood the concepts, and from a business use case level I gradually progressed towards writing my dissertation on AI use cases, what values it brings to a business and how firms are working to implement AI."
Prasana then joined Gamesys as a Senior Product Manager which involved working on projects that utilized Big Data, Machine Learning, Cloud technology and AI. He worked closely with the Gamesys Head of Data on Marketing and organisational aspects of each project as it was important for the organisation to understand the value of the proposed solution and the benefits offered to stakeholders. These AI/ML use cases involved working closely alongside Google, who provided a great deal of platform, knowledge and support.
His technical expertise and engineering background enabled Prasana substantially when working on these projects as he already had a good understanding of technological architecture. Armed with an Engineering Degree and an MBA, Prasana formed an intimate understanding of both statistics and data that was necessary when delivering on the projects he led.
What are the differences between a standalone product and one that incorporates AI in the platform?
When queried about the differences between standalone products and Products that have a fundamental incorporation of AI, Prasana commented: "AI sits behind the screen, in most cases." He went on to explain that typical standalone products require making a start from a contemporary product management stance “you start from understanding the problem context, the data (Volume, Veracity, value and velocity) associated to create models, new data points/sources (if needed), ways to make the data available to train the model, strategy to ship the model to production by creating a product loop and maintaining it towards enhancing business value. It's an iterative process as the data context shifts, market shifts and as well the choice of model when above happens."
In terms of products that incorporate AI within platforms, it's important to start with an initial understanding of what the problem area is. This means having knowledge of the life cycle, the organisation's business life cycle, alongside an appreciation of the data landscape. Prasana typically works through a framework for identification of problem areas, known as the hierarchy of needs. He aims to reach an understanding of data elements used by the business, the types of internal data transactions, and the details of external data transactions. When talking with stakeholders, he also needs to discover other areas which require focus. In essence, creating the AI-based platform sits behind understanding data landscape planning and Product Management.
How data-oriented do you have to be with AI products?
From a Data Engineering or Data Scientist's point of view, as well as from a Business perspective, it's important to provide a predictive-based election, as this enhances the product's value. The entire process is based on data, whether that is predictive or historic. It takes a good deal of time for models to mature and create the Product loop that's required.
It's not essential for people working in this field to have technical expertise. What is important is that they possess a firm understanding of the Product loop and of data, alongside a business appreciation of the way machine learning models are trained and how to get them into production.
Once a product enters the production phase, there are only two considerations. These are ‘precision’ and ‘recall’. Within the arena of precision, there is a limited amount of acceptable false positives, so teams need to be highly precise in order to meet expectations and regulatory requirements.
Mature organisations tend to operate a centre of excellence for projects of this nature, and data scientists and engineers form the backbone of these departments and generally take responsibility for their own products.
Prasana states that the timeline is vital. This sort of technology can shape itself and learn, but it's important to have a knowledge and appreciation of strengths in order to provide the data to the model. Within the model itself, there is no fixed timeline for development. However, there is always a probability of failure. The role is a mixture of science and art. The scientific aspect entails feeding the data into the product, while the art aspect is finding the right prediction and ensuring management teams are happy with the output.
So, talking with all business stakeholders to achieve an understanding of likely problems is really critical to the development process.
Collaborating with teams
Prasana goes on to explain to John that in an ideal scenario, he would arrange at least one workshop session with stakeholder teams, as these help immensely and can provide more value for the Product. Product teams and Consumer teams also form a major role in the development process, alongside other organisational and external stakeholders, so it's vital to ensure high levels of collaboration to achieve the required understanding.
Longer-term thinking about products and iterating on products
In the longer term, Concept drift causes problems because the predictions become less accurate as time passes. In machine learning, the concept drift means that the statistical properties of the target variable, which the model is trying to predict, change over time in unforeseen ways. Adding more data points will expand the abilities of models. Constant iteration is involved within the role of the Data Science team in order to move the model to finalised status.
Prasana added that agility around data points and organisation is vital within his role. Any discussion surrounding data points needs to take into account Government regulations concerning that data and the legitimacy of processing consumer information. This can add a completely different perspective to AI, as data government and stewardship is crucial. Many businesses ignore data government and the data landscape, and it is an extremely technical and difficult aspect of the role involving ethics and the sort of human bias that can be built into any model. It's important to conduct a bias minimum checkpoint at the time a model is put into production.
Advice for working with AI products
The specialist Product Management Recruitment Consultants at Consortia have seen a growing number of candidates keen to move into the AI space within our specialist product management recruitment agency. Prasana offers the following advice to anybody in this position: "You have to understand how the data world works and look at a broader perspective than just the data analytics aspect to product management. It's more about knowing data points and what the company offers that adds value to the business." He adds there are a growing number of online courses providing an understanding of data science and a lot of valuable open content on the web.
Do you have an AI Product Management role that you need to fill? Why not reach out to our specialist Product Management Recruitment Consultants who can help you to build your team accordingly. You can contact us here.
Are you looking to build your Product Management team to carry on with your growth plans? We can help you to scale teams accordingly, working to the brief and advising throughout the hiring process. As a specialist Product Management recruitment agency, this is what we do. Contact us here to send us a job brief.
Thank you to Prasana Murthi of Receipt Bank for taking part in our Industry Insiders series. While you are here, why not check out some of our other Product Management insights...
- Why being a product manager is harder than being a product leader by William Rowe of Just3Things
Being a Product Manager is tough, but is it tougher than being a Product Leader? At first sight, the obvious answer is no. Juggling a leadership role, plus needing to use his/her Product skills must be tougher? William Rowe, though, has a different opinion. As an expert Product Management Recruitment Agency, our tenacious Product Recruiter, John Magani caught up with William about why he believes this is the case.
- What separates the top 1% of product managers from the top 10%? by Pouya Jamshidiat at Lloyds Bank
Pouya Jamshidiat has a passion for entrepreneurship, innovation and design thinking. For someone who has designed, managed and owned products for global leaders in finance and technology — Lloyds and IBM to name but a few — such interests are essential to success.comments powered by Disqus