Four Actions To Keep Your AI Product Strategy On Track

Four Actions To Keep Your AI Product Strategy On Track

In addition to keeping track of AI developments, chief product officers (CPOs) also have a day job, which involves articulating the vision for their product portfolios, holding product managers accountable for getting new functionality rolled out and listening to customer requirements. As CEOs demand monthly updates on how AI will boost the firm’s valuation and engage customers, what should be on the CPO action list?

#1 Push past scepticism and focus on the big bets.
As with all prior waves of technology innovation, the initial excitement about generative AI (GenAI) now faces a backlash. Enterprise users point to latency in chatbot responses, information security concerns, court cases about LLMs based on copyrighted content, AI hallucinations, excessive power consumption, high costs per prompt, insufficient GPUs and a lack of reliability. Amidst this rising wave of AI angst, many people have forgotten that ChatGPT 3.5 is the fastest-growing consumer software application in history. The GenAI-powered chatbot reached 100 million users within two months. To keep AI strategy on track, CPOs need to educate executives about the billions of dollars in investment that Microsoft/OpenAI, Alphabet/Google and Meta are putting into next generation LLMs, which will improve the performance of AI applications in line with standard enterprise expectations.

#2 Implement a retrieval augmented generation model to demonstrate success.
As one of the prime beneficiaries of the flood of interest in GenAI, computer hardware manufacturer Nvidia has a major stake in ensuring that enthusiasm is translated into business investment. To get business leaders on side, it has supported the concept of retrieval augmented generation (RAG), which was first defined by AI researchers in a study published in 2020. The underlying idea is that by connecting generic LLMs – such as Google’s Gemini – with specific, curated information, AI vendors can overcome some of the barriers raised by business users such as a lack of attribution, confidential data leakage and the presentation of invented facts to users (hallucinations). For example, a RAG model could apply the query interface and reasoning power of an LLM with the EU Green Taxonomy and corporate revenue data to provide a detailed assessment of alignment with the taxonomy. The time-saving potential is enormous.

#3 Educate the executive team on the future of multi-modal agents.
For CPOs who are keen to secure funding for their AI product strategy, it is essential to keep the Board updated on what the AI software landscape will look like in three years’ time. This means shifting the mindset from text in/text out AI chatbots to multi-modal agents. Service providers like EY are focusing their AI enterprise implementation services on agents. The reason? LLMs are able to transform inputs in one language – such as French – into another language, like Python code, which can instruct another software application to produce an image, music or video. This multi-language transformation capability is central to changing AI from being a sustaining innovation to becoming a force of creative destruction. Multi-modal agents will be able to orchestrate multi-step business processes without a human in the loop.

#4 Explain how GenAI will create a tectonic shift in software value propositions.
For 30 years, the value proposition of enterprise software has been anchored by the promise of improving worker productivity. From the green screen days of ERP to cloud CRM, the paradigm has been based on developing human/computer user interfaces that improve process execution, data management and worker productivity. Sarah Tavel, a general partner at Benchmark Capital, has eloquently made the case that in the near future, enterprise applications that leverage LLMs and multi-modal agents will be able to execute entire processes. We will move from AI-in-the-loop to human-in-the-loop to autonomous systems. The implications? The economics of enterprise software will change completely. The role of software will be to perform as many actions as possible in a business process with zero human intervention. Software will not be sold on a per-seat basis because there won’t be hundreds of humans logging in. The idea of a user experience will disappear. The faster execs understand the extent of the change, the faster they will protect their business’s value by funding AI product strategies.