In today’s tech landscape, Generative AI (GenAI) emerges as a pivotal force, offering businesses the opportunity to pioneer products that stand at the forefront of their respective industries, driving success at the product level. Let’s delve into actionable steps for implementation and explore the potential applications of GenAI in your product development journey.
OpenAI and their flagship product ChatGPT disrupted and expedited the rise of AI which was evident by a huge rise in their user base in just 2 months after launch. While ChatGPT has taken most of the limelight, there is more to GenAI. Unlike traditional predictive models, GenAI stands out by prioritizing data generation over outcome prediction. For instance, natural language processing (NLP) algorithms, a type of AI that predicts text sentiments, and GenAI looks to generate the next word in a sentence leading to creative yet sensible outcomes.
GenAI models are exceptionally versatile, and able to support multiple tasks that are not possible for traditional predictive models. Ranging from video creation with the likes of tools called Synthesia or to generate images with a single prompt like Adobe Firefly, these models are brimming with potential to drive significant impact across numerous industries. ChatGPT a prime example of GenAI is a large language model (LLM) that has been trained on a wide range of massive datasets which allows it to be equipped with the ability to generate blogs, fix and write code, generate recommendations for best practices for any kind of workflow, allowing for tangible, real-world impact. It also underscores GenAI’s acuity and resourcefulness.
A recent study by Gartner predicts that by 2028, companies expect to replace 60% of their work tools with those powered by GenAI. However, it is worth noting that there are widespread concerns around GenAI models’ tendency to generate inaccurate information or “hallucinate”. Are LLMs useful in the current state to augment or replace valuable tasks or do they need to be trained on specific things to extract usefulness? And by extension, is GenAI the next advancement that will change the world or just hype?
There has to be a path to navigate the debates, emotions, and exaggerated claims to have a clearer understanding of the impact AI will have in the future? And finally, Can your product benefit from AI and should you define a GenAI strategy for your product-led success?
Let’s try to demystify the buzz and find a rubric to evaluate our next move in the world of AI.
Common ways companies are integrating AI into their product
It is difficult to stay abreast of the ongoing news about AI, and it seems that almost all of the days, the media announces to the world about a “new AI product” that is launched by a particular company. Though not all of it can be used, how AI might create an environment for us to design products has to start by classifying all of them by their idea.
The Easy Button Approach: Adding AI with ChatGPT’s API
Don’t rush to shove AI into your product just because everyone else is. While slapping on a basic chatbot might seem easy, it’s not always the smartest move. Here’s why: Basic chatbots (bronze tier!) don’t understand your product, they just mimic ChatGPT. It is easy to implement and lets you test user interest for a smarter chatbot that could be built out ultimately. Users get ChatGPT-like access without switching tabs. This approach is a good starting point for big companies (think Snapchat) and individual developers to dip their toes in AI without a huge commitment.
Fine-Tuning for a Competitive Edge: Training AI on Your Expertise
Now, to extract a real problem-solving use case for a company with AI, big companies are able to leverage their rich datasets. Instead of using existing models as is, they can “fine-tune” it with their own industry-specific data. Having access to company data, allows this AI to get better at solving company-specific problems that are relevant to the business. Giving it an edge over the other foundational models, and feeding high-quality data to the AI is definitely a smarter way to leverage existing products and give them a competitive edge.
Skip the API, Train for Advantage
Does the competition cause you to think that you need to add AI to your product even if it isn’t needed? Hold on! Remember the core principles of product roadmapping. Don’t do a hasty implementation of AI and think of it as extras. Make sure it aligns perfectly with your product’s overall identity and user experience. This integration of AI does sound like a smart way, but the “how” is important. Here’s an approach to guide you through the decision to incorporate AI and, if you choose to proceed, how to do it effectively.
Should You Incorporate AI Into Your Product?
Successful products always start at solving customer problems and driving customer value.
A good place is to start to leverage qualitative and quantitative research to understand the kinds of problems users run into time and again. Capturing the needs across different stages of the funnel, i.e., acquisition, activation, retention and sorting them in the order of the friction and pain users’ experience will lead you with a few options that can help the users become more successful on the platform.
However, solving pain points is never enough because bets should align with business outcomes. Is the product adding horizontal features to drive additional personas to be successful? Which product areas are most critical for users to be successful in and are they stuck currently?
The key to unlocking true value lies in finding the intersection between what solves user problems and what drives business success.
One additional nudge for my fellow PMs here is to consider if adding AI is still in the scope of the expected use cases for your product. Most mainstream use of AI is in the form of a chatbot, but is chatting a primary way of using your product? It is important to consider the use case along with the user journey. Tacking on the chatbot interface is not likely to drive any benefits if the product is not chat-driven. Eg: Adding AI to WhatsApp or Zendesk could most likely enhance the experience since primary interactions on the product are chat-based.
On the other hand, if primary interactions are not chat-based (like Miro or Coda), a little more thought has to be put into understanding seamless integrations of AI rather than shoehorning chatbots into the product.
This is particularly vital because features are never free. Surface expansion beyond usual usage will bring in additional tech debt in order to maintain the surface going forward.
The final step with the prioritized list can be assessed with a risk vs demand assessment. How bad would it be if wrong information spread? (risk) And, “Do people really need this kind of stuff, not just because it’s popular now?” (demand). It is all about designing product delight and defensibility to make up for the costs of implementing AI such that the end experience is improved and is more user-friendly. The low-risk, high-demand projects are a great place to start. A good example is using GenAI to summarize transcripts generated by video recording tools like Zoom. It is all about thinking of ways to build defensibility with AI and bring user-friendly change to the product.
The future of work is here, and AI is shaking things up! Even in my daily tasks, I’m seeing the impact. As a product manager, writing tons of repetitive SQL code used to be the norm. But now, AI helps me automate that, freeing up my time.
The point? Even if you’re not ready to jump into AI for your product yet, get started! Explore AI tools, understand the new user experiences they create, and brainstorm how they can benefit your organization. Don’t get left behind – the future of work is AI-powered, and it’s time to embrace it!
