In this #mtpcon San Francisco talk, Aniket Deosthali, former Chief Product Officer of Walmart, dives into the potential that AI brings for product managers and the challenges that lie ahead of building AI-driven products. Watch the video to see the talk in full, or read on for an overview of the key points:
- AI is a sustaining innovation
- Consideration-context framework
- Prioritize truth and training to leverage AI/ML
- Five steps to AI-driven product-market fit
AI is a sustaining innovation.
First, some basics about ML. It is two-step-oriented:
- Training: is the process where curated data is given to an AI model to learn and make predictions.
- Inference: is the process where the pre-trained algorithm is ready to answer questions in real-time.
The whole process summarizes one thing: data is far and away the most important component of machine learning. Existing companies already have a massive amount of data from consumers and retention in their possession that can be used by AI models to make better predictions. As a result, Aniket believes AI is a sustaining innovation as it helps existing companies.
Consideration-context framework
Within this framework, on the Y-axis, you have consideration. Consideration is the amount of effort it takes to make a decision. And on the X-axis, there is context, which is the volume of data needed for a machine-learning model to make an effective prediction.
The benefit of using this framework is that it helps product developers identify the areas in their current market landscape where their product can gain significant success.
Prioritize truth and training to leverage AI and ML.
No matter how beneficial AI is, it comes with its own set of disadvantages. AI is a probabilistic system, which can be an issue if you want to provide your users with a stable user experience.
The best way to tackle this problem is to build a hybrid model where humans work alongside AI. This could include feeding AI thousands of hours of data and using reinforcement learning with humans to provide AI models with feedback on what good looks like.
5 steps to AI-driven product-market Fit
Step 1: Define and prioritize your use cases
Whenever you think of building a product, the easiest thing to do is think about the solution. However, Aniket believes the problem deserves 100% of your attention.
With the help of use case maps, you can pinpoint the key requirements of your product, which might include:
- What is the problem you are trying to solve?
- Who is your target audience?
- Is there an alternative solution available?
- How often does the problem occur?
The benefit of using use case maps is that they not only help you identify the most common and important problems your target audience is facing but also avoid the probabilistic disadvantage of AI models.
Step 2: Formulate a 10-x product hypothesis
The five key elements of this product hypothesis include:
- Proactivity: Can you anticipate the customer’s needs?
- Personalization: Can you provide relevant information that meets the requirements of your customer?
- Personality: Can you infuse a distinct tone that connects with your target audience?
- Automation: Can you build automation features to reduce manual work?
- Accessibility: Can you take advantage of existing platforms to make your product more accessible to your customer?
Step 3: Check your riskiest assumption
It’s time to test your hypothesis. You can test it against three broad buckets of risk, including:
- Customer risk: Ensures whether your product is actually solving a problem your customer is facing and whether it offers a better solution compared to the pre-existing products in the market. It also helps identify any biases in your product. Because AI is a probabilistic system, it can’t offer a consistent user experience. Addressing the biases ensures a seamless user experience for the customers.
- Business risk: Addresses the financial aspects of your product, such as whether it will be a profitable venture and whether you have enough funding for machine learning operations.
- Tech risk: Offers answers to questions like whether the product is feasible in terms of hardware and software and is there enough data available for models to make effective predictions.
Step 4: Rapid prototyping
When thinking about prototypes, the first thing that should come to mind is to build something that generates enthusiastic and positive feedback.
With that in mind, Aniket’s approach here is to use wizard-of-oz prototypes that allow you to experiment with your concept cost-effectively. The main aim here is to figure out something that provides value to your customers and understand how they interact with it before investing in a full-fledged implementation.
Step 5: Launch your minimum-lovable product.
To conclude, Aniket advises using end-to-end solutions and open-source tools. With the rapid advancement of technology, the infrastructure behind these tools is rapidly growing. As a result, you can streamline the process of building use cases and basic applications without the help of a large developer team.
The key takeaway from this talk is that if you can create a product that perfectly fits your market’s needs, you can take the advances of modern technology and build long-term value for your company.
Want to turn the learning from this talk into action?
Our exciting #mtpcon San Francisco 2023 Keynote Kit brings you all of the insights from our San Francisco keynote talks plus additional helpful discussion points and thought starters so you can easily translate the insights into actions – perfect for making effective improvements to your role, team, and product. Plus, get an email notification when each talk is published! Sign up now.
Comments 0