AI for Product Teams
Coders and Product managers are two of the areas most ripe to be transformed through comprehensive adoption of AI. Jobs will change, it will be important to migrate your talents to where AI drives them.
Over the last 30 years or so, the number of coders has grown dramatically to accommodate professional needs. Starting below a million in the US in the early 90’s it is estimated there are well over 30 million professional software engineers as we head into 2025. That count does not include the millions and millions of web development tool users managing their own needs, with little formal coding training, relying on tools such as WordPress, HubSpot, Spotify, GoDaddy, AWS to generate the templated code that is needed.
For anyone who has used AI coding tools like CoPilot from GitHub, it is easy to see that AI tools thrive generating code. They are largely semantic language engines after all. Given most coding languages are meant to be semantically unambiguous for a computer to execute the code properly, the sophistication AI embodies to understand and generate ambiguous spoken languages like English, is largely left unneeded. Code generating tools still suffer from garbage-in/garbage out risks (as do AI chat tools like ChatGPT). This is where AI-augmented skills for human operators (you and me) become critical, to get the value you want to realize, and possibly, to preserve the jobs.
For Product managers, the essence of the Product role is the synthesis of streams of requirements (input) to create the output an Engineering team can use to economically build, and a business can take to market to generate revenue. The more unambiguous and consistent the output a Product team can be produce, the more likely coders and sales teams will be able to meet the needs identified. While there is a general risk of homogenization of thought and approach as we become dependent on AI (as there was with spreadsheets in Finance long ago) – the benefit for Product is alignment, consistency and completeness of analysis from the generated artifacts produced over time.
Growing the percentage of time allocated to innovation and market differentiation in today’s pre-AI model, comes down to what is in the stream of input being synthesized and presented to the emerging AI tools. Product role users must learn how to carefully craft inputs (a.k.a. prompts) to make sure AI is challenged to compute past conventional analysis (e.g. show some thinking) and offer back diverse insights the human factor can affirm and lock in for execution. It is less clear whether AI will cause elimination of Product roles, but it will certainly change the way in which they are carried out, and by inference, a positive shift can occur in terms of how the Product role contributes to the success of the business.
The following sections outline five of the key phases to work through as part of onboarding AI as a critical element in the processes and workflows of the Product function (and to some extent, the Engineering functions as well).
Measuring Impact of AI
Winning business and product strategies and their annualized plans always include measured goals. Whether they are in the form of KPIs, OKRs, Balanced Scorecards or other measurement frameworks, those that win understand that measuring your progress towards goals help guide a team forward, in an aligned way.
AI carries with it a lot of promise and a lot of unknowns at this stage of its maturity. AI is also evolving at the most rapid rate compared to any prior transformational technology. It may take decades to fully settle the impact AI will have on businesses, roles and day to day society. As such, it will be easy to start out with one set of goals for your team and quickly realize they are not the ones you should have set out as your understanding of AI deepens. Goals need to be assessed and reassessed as more is learned. Measuring goals needs to be refined so they remain meaningful. It would not be outlandish to measure the efficacy of your goal setting keeping an eye on how much flex they go through. The goal for measuring the goals would be to reduce flex as a way to affirm your own understanding of what benefits your business goals and expectations are solidifying.
The following provides a high-level view of critical questions to ask about your business to help identify what goals you should have, how they would be measured, and most importantly, how your team will react if the measures show a goal is off track.
Faster time to market It is unclear if AI will improve, in material ways, the time to market for product initiatives. The coding tasks can be accelerated with AI but coding is, while a major piece, still not the only activity to bring new things to market. The Product function, which is supposed to work ahead of the active Sprint cycles, does need to pace itself along side Sprint cycle timing.