An Intro to AI Powered Product Development

If you look around the world, everyday life seems to be influenced by products of all kinds. Whether it’s entertainment, health care, or necessities of life like food and shelter, everything appears to be affected by products developed, manufactured, and sold by different companies. Let’s use some examples here to show how various products serve and delight us.

Take something like the automobile, which has been around for a long time. It takes us from point A to B safely. Along the way, it has to provide us with comfort and pleasure: we can listen to music or other audio content; we can talk to our friends and family while keeping our favorite beverage from spilling.

Buying a cup of coffee at Starbucks has been designed and developed to give us a pleasing experience in ways we don’t realize. There are many other examples in categories such as medical products, food, and furniture, where products have the intended functionality and the more complex user experience in mind.

Maybe I am stating the obvious. But, developing and launching products that can stay successful is difficult, not just because of the complex and fickle customer, but also due to the various factors that influence the destiny of a product’s life — supply chain, government regulations, competitors, and many others. Companies worldwide spend a lot of human and financial resources to get their products to the market. The global product development services industry was close to $8 billion in 2020.

Current Product development techniques

Companies worldwide employ various organizational structures, focused expertise, and techniques for their product development. Many functions such as Research and Development (R&D), Industrial Design, Marketing, Engineering and Manufacturing work together to ideate and bring those ideas to market.

These teams use skills and tools like Voice of the Customer(VOC), Kano analysis, Design for Six Sigma(DFSS), Advanced Product Quality Process(APQP), and a wide range of experienced technical experts depending on the kind of product. In many cases, we can outsource these experts and skills.

Typically, experience with previous development and other forms of tribal knowledge play a significant role in the various decisions needed in the product journey. In essence, successful product development depends on making optimum decisions at every stage.

Sometimes, poor choices can be extremely costly. If you google “Epic product launch failures”, several examples will come up. Products fail to be successful for many reasons: not understanding customer needs, price, competition, or simply bad timing. It has to work at every stage in its journey: from creation to consumer.

For example, a design could be extremely popular with end-users at a specific price, but it will die a sure and swift death if there are no profits. These roadblocks are not limited to new launches. Products that were successful when they first came out face similar challenges to continue in a positive direction. Consumer trends, new government regulations, and competitors with cheaper and better models have to be monitored and overcome. This type of monitoring comes at a cost. To this end, companies large and small frequently employ product managers, brand managers, engineers, and regulatory compliance administrators.

But let us take a step back and see what all these people and resources involved with product development are doing.

Essentially, product development happens along a value chain — different functions and activities such as R&D, design, marketing, engineering, supply chain, and manufacturing and logistics form various links. The product progresses from idea to launch and distribution along this chain.

Big and small decisions are being made continuously by taking data from memories, test reports, images, videos, and customer surveys and applying some best practices. This best practice comes from experience, a more structured technique or process, as mentioned earlier, or usually a combination of both.

How can AI help?

With the rise of the internet, information and data are ubiquitous now. Companies have large amounts of unstructured data in documents, images, and videos that potentially contain valuable information pertinent to a product under development or an existing one.

Also, most companies have experienced personnel with specific tribal knowledge at each link in the value chain, or they might be outsourcing this function. Here is where AI tools could come in. Is it possible to take the data from various sources mentioned above and create algorithms that mimic best practices, tribal knowledge, and product development experience to launch and sustain successful products? I think it is. Let me elaborate with a few examples at different parts of the product development value chain.

Product Ideation, Research, and Monitoring

At this stage, it is all about general brainstorming and the feasibility of an idea for a new product. Some questions to be answered are — will it sell? Is it the right price? Who is my competition, and what are the general market trends? What are some of the supply chain challenges? Also, can I trademark and patent my idea?

If it is a product that is already on the market, there are other challenges. For example, are my competitors catching up? Are the market conditions changing?

Let us look at available data and the best practices and the AI tools to combine both to provide the maximum value at this link in the product development value chain.

Data: The available data is pretty much everything on the internet, both open-sourced or paid outlets and internal sources in various forms like documents, videos, and images. Some useful sources of data for product development and innovation include ArXiV, USPTO (Patents) and People Data Labs.

Best Practices: The knowledge and experience to categorize and sift the available data into helpful categories such as competitive intelligence, market trends, customer sentiment, government regulatory challenges, Intellectual property, and supply chain. The ability to also recognize and dive deeper into valuable data points is crucial.

AI tools: Techniques such as Natural Language Processing (NLP), Machine learning-based Image recognition, and Video processing could potentially sift through and categorize all the data according to the best practices mentioned above. Using tools such as a cognitive search, related and valuable concepts could be suggested. Target customer and market sentiment related to the ideas or products could be displayed using NLP-based sentiment analysis.

Here is an example of how consumer sentiment could be tracked and displayed for a product idea:

Companies such as Patsnap and Quid use NLP and ML algorithms with patent databases, research papers and other data sets to extract useful insights related to product innovation.

Product Design and Engineering

Here is where you know what product you are developing but need to get into the details. What are the customer needs? What are the best features, and what materials to use? How do we manufacture it, and what is it going to cost? Finally, how do we test it?

Once again, let us look at the available data, required best practices, and potential AI-based tools to produce value at this level.

Data: There is no change here. The available information is the same as before — the internet and related sources and internal company data, but the internal source could be different; maybe it is more engineering-centric than commercial. All of this data could be scattered and unstructured. Internal data could be used to train ML models or the creation of data sets could be outsourced cost-effectively. Sama is an example of a company that provides such services.

There are various free sources of data available on the internet ranging from official government data to open-sourced data from sites such as kaggle.

Best Practices: The knowledge and experience to understand customers and stakeholders, their needs from this product, and how to test the final product for these needs. The ability to identify the optimum design features based on various requirements and also manufacturing methods is critical. Assessing patentability or other intellectual property considerations at this stage is also essential.

AI tools: NLP can play a vital role at this stage again. Sifting through available data and categorizing and displaying it to help identify product stakeholders, requirements and brainstorming potential test methods is done through carefully adjusted NLP algorithms that leverage generic or customized Word2vec models. The same algorithms could nudge you towards data that helps identify optimum design features.

These features are evaluated for patentability or if they violate any other existing patents. Here is a cool video showing how AI can be used in product design. Platforms such as RiMo use real world data from previous drilling projects with ML algorithms to help advise and design future projects.

The above examples merely illustrate the potential power of AI for product development — both for new products and successfully sustaining existing ones.

At FortuitApps, we have been developing a tool that aims to do what this article describes. Devtom is an AI-powered platform for product development, connecting millions of distributed data sources with AI algorithms to provide much-needed a-ha moments at every stage of the product development value chain. In subsequent articles, we will do a deeper dive into how AI can affect and help new product development with more examples and details.

Also published here: https://www.fortuitapps.com/fortuit/techproducts.html