Exploring Strategy For AI Products
Understanding Why AI Products Require Companies To Rethink Traditional Product Strategy
Strategy is a critical part of how companies build great products.
It guides their product decisions at every stage, such as the customers they target, the markets they compete in, the problems they solve, etc.
Major technology companies have established frameworks for successful product strategy. In most industries, there is a good understanding of what the path to success might look like. However, innovations disrupt conventional approaches and often force companies to rethink their strategy.
Artificial intelligence (AI) is the latest transformative technology that companies want to capitalize on. While AI promises great benefits to both businesses creating products and the ones using them, it also introduces additional complexity to product strategy.
AI products are fundamentally different from traditional software products. They solve people’s problems with an incredible level of personalization and flexibility, which presents new challenges and opportunities.
I decided to explore how companies might adapt their strategy to build successful AI products.
What I Will Cover
What Is Strategy - Defining what your goal is and how you will get there.
Why Do AI Products Need A Different Strategy - AI products operate in fundamentally different ways from traditional software products.
What’s Different About AI Product Strategy - AI products have to deal with different problems, requirements, and challenges.
Imagining An AI Product Strategy
Takeaways For Product Teams
Conclusion
What Is Strategy
A strategy defines what your goal is and how you will get there. It helps you to course-correct along the way and accomplish your goal even if the path to get there changes. It has to be logical, feasible, and actionable.
What Is Product Strategy?
Product strategy defines what your product will accomplish, how you will do it, and why you will succeed. It is rooted in your organization’s mission, vision, and business goals. It guides all your actions and decisions, even as your customer and business needs change.
Organizations can either create a product strategy framework or adapt established frameworks from successful companies. However, there is no perfect framework because what works for someone else might not work for you. You just need a systematic way to establish, evaluate, and execute your goals. Something that makes sense for your organization and your product.
Why Do AI Products Need A Different Strategy
AI products operate in fundamentally different ways from traditional software products.
Traditional Software Products
How They Work
Traditional software products follow predetermined rules and a consistent sequence of steps to help people complete tasks. For example, to complete a certain task the software will always do steps 1-2-3 (in that specific order).
Every single step the product takes needs to be programmed.
How People Use Them
Typically, people engage with these products in a limited number of ways. The product teams design the product to facilitate certain well-defined interactions. These interactions are based on the sequence of steps the product has to complete to produce a useful and valuable output.
How They Evolve
The product capabilities are incrementally expanded over time as product teams add the ability to perform other tasks (with their own specific sequence of steps).
AI Products
How They Work
The main difference with AI products is how they complete tasks. These products use AI models to determine the necessary sequence of steps to complete a task. For example, to complete a certain task the software might choose to do steps 1-2-3, 4-5-6, 7-8-9, etc., depending on the situation.
Every single step the product takes does not need to be programmed.
How People Use Them
People might engage with the product in many different ways, and the AI needs to adapt as necessary. The product teams have to account for a range of interactions based on people’s specific needs and context, while still producing a useful and valuable output.
How They Evolve
While the AI product capabilities are expanded over time (just like traditional products), these products can learn to accomplish new tasks in new ways without additional programming. Their ability to “evolve” is what makes them fundamentally different.
What’s Different About AI Product Strategy
AI products have to deal with different problems, requirements, and challenges.
Defining An AI Problem
AI product strategy focuses on a problem that AI is best suited to solve.
Product teams need to define a problem that is valuable to the customer and the business, which their AI solution can solve. The nature and complexity of the problem will determine the requirements for implementing an AI solution.
Establishing AI Implementation Requirements
AI product strategy considers the data, talent, and technology needed for effective AI implementation.
Data
Product teams must uncover the right data and determine how to use it effectively. AI models require high-quality information to produce a valuable output because their performance is only as good as the data they learn from. Therefore, product teams need robust data strategies to collect, store, and manage the necessary data for AI implementation.
Talent
Product teams need specialized expertise to develop, refine, and implement AI. Even if they use AI models from other organizations (OpenAI, Google, etc.), they still need experts to tailor the AI for their target applications. However, the necessary talent is often scarce and in high demand. Therefore, product teams need to consider the availability of AI talent in their strategy,
Technology
Product teams need the technology infrastructure to process information and deliver AI outputs fast enough to be truly useful. The necessary computing resources can get quite expensive because AI Models need a lot of computing power to complete complex tasks. Therefore, product teams need to understand their computing requirements to manage the operational costs of AI.
Understanding AI-Specific Challenges
AI product strategy accounts for the unique challenges involved with AI.
Execution
Since AI products have dynamic outputs, execution is not always consistent. While great product strategy is important, the user experience and business model give people a compelling reason to choose a product. Therefore, product teams must closely monitor their AI outputs and interactions to identify execution issues, improve their product, and meet people’s expectations.
Oversight
AI products can produce dynamic outputs based on people’s specific needs and requirements. While these are personalized outputs, they are not exactly predictable. The AI uses different logic than us to understand information, which leads to unexpected, inaccurate, or misleading results sometimes. Therefore, product teams must implement systems to verify and oversee AI outputs.
Accuracy
AI products have varying levels of accuracy. In critical tasks (healthcare, law, etc.) high accuracy is essential because the outputs impact people’s lives. However, improving accuracy can require additional training and refinement, which increases research and development costs. Therefore, product teams need to understand the accuracy required for their desired applications.
Explainability
AI products can often turn into "black boxes" where inputs mysteriously turn into outputs. However, teams need to know how the AI arrived at a specific result to diagnose issues, such as using incorrect logic, biased data, or incomplete information. Therefore, product teams need to focus on explainability to improve the accuracy, validity, and quality of their results.
Security and Privacy
AI products might have to go beyond the standard data security and data privacy requirements because of their dynamic nature. If the AI samples protected information (Personal Identifiable Information), this information must not be exposed to unauthorized users when generating outputs. Therefore, product teams need to take measures to enhance security and privacy.
Regulations and Ethics
AI products in certain domains (healthcare, finance, law, etc.) might need to comply with rules and regulations. Therefore, product teams need to include a plan to meet standards and requirements in their strategy. Furthermore, beyond the regulations, they also need to consider if their product is operating ethically in the best interest of the people who use it and society in general.
Imagining An AI Product Strategy
The most well-known AI product today is OpenAI’s ChatGPT. While companies have been using some form of AI for a long time, the concept of providing AI as a core service is somewhat new, especially to consumers. Therefore, the strategy driving AI products is still evolving. It might be too soon to tell what the right strategy might be.
However, we can imagine what the process of developing an AI product strategy might look like by looking at conventional product strategy through the lens of AI.
While the exact steps might vary between organizations, we can broadly categorize them as follows:
Establish A Vision: What do you want to accomplish with AI? What are the specific goals?
Understand The Customer: Who is your target customer? What are their needs, motivations, and expectations? What problems can you solve for them?
Define An AI Problem: What problem are you trying to solve with AI? Why does this require an AI solution? Why your AI solution? Which applications have the strongest business case for AI?
Identify The Opportunities: What solutions are the customers currently using? Where are they lacking? What opportunities can you address? What can you do differently?
Establish AI Requirements: What type of is data available for an AI solution? What expertise will you need? What infrastructure will you need to produce useful AI outputs?
Develop A Solution: How will AI solve your customers’ problems? What AI solutions are feasible? How likely are they to succeed?
Address AI-Specific Challenges: What are the potential problems, challenges, and risks? How will you manage them? What are your core assumptions?
Align the team: What is your team’s shared understanding of the customer, problem, and solution? Does everyone agree on what needs to be done and why?
Create A Roadmap: What work will be done? How will it be managed? What are the priorities? What is the timeline? Who will be responsible?
Establish Success Metrics: What does success mean for your product? How will you measure it? What benchmarks will you use for validation and course correction?
Execute And Refine: Implement the strategy, measure success, and adapt as needed.
Companies Implementing Distinct AI Product Strategies
The recent reveal of OpenAI’s GPT-4o and Google’s Project Astra signaled a clear shift towards making AI products more immersive. These products allow users to interact with AI in multiple formats (text, speech, video, etc), which makes the interactions with AI feel more natural.
Furthermore, the Large Language Models (LLMs) made by these companies are the foundational solutions that power AI applications made by other companies. Therefore, AI products that use these LLMs will also be able to deliver these immersive experiences.
These companies have a clear focus on the human experience, and their strategy for making AI products that seamlessly fit into our lives will ultimately make AI even more useful to people.
Takeaways For Product Teams
Recognize why AI products are fundamentally different from traditional software.
Evaluate your customers' needs, motivations, and expectations in the AI context.
Assess potential AI applications based on the data, talent, and technology needs.
Focus on identifying a complex customer problem that AI is best suited to solve.
Address the unique AI implementation challenges for your target application.
Understand the core technologies that are enabling your AI solutions.
Conclusion
The more problems a product can solve for people, the more valuable it is. AI’s unique capabilities position it as a single solution for many (if not all) of our needs. While AI products have the potential for massive commercial success, their dynamic nature requires a strategic shift in how companies approach product development.
Product teams need a shared understanding of the AI’s capabilities (what it can do), applications (where it can be used), and limitations (what it cannot do). A more deliberate and targeted strategy ensures that AI will be used effectively to solve the right problems, build the right solutions, and create the right outcomes.
While there are a massive number of AI companies right now, only companies with clear and compelling AI product strategies are likely to thrive in the long term.
Thanks For Reading!
If you haven’t already, please consider subscribing and sharing this newsletter with a friend.
I hope you have a great week!
References
Product Strategy
AI Product Strategy