In the playtime stage, use AI more as a thought partner and less as a tool, advises Jan Bosch.
As companies start to explore what AI and especially LLMs and agentic AI can do for them, the first step is what we refer to as “playtime.” The term may sound a little unserious, but the idea is that the focus is on individuals starting to use AI for personal productivity. This means that the tasks employees used to conduct by hand are now supported or fully automated by AI. Examples include generating response emails, automatically generating documents or summarizing material sent to them.
Although the benefits for the organization aren’t as significant as the later steps in the maturity model we’re presenting in this series, the productivity improvement for individuals can be quite significant. Based on our interview study and broader experience, we can recognize several types of AI-based automation that people use in this stage: automating routine cognitive tasks, managing information overload, improved decision-making, sparring partner and task execution assistant.
Looking at automating routine cognitive tasks, LLMs are especially useful in analyzing and summarizing long documents and emails. Simply getting accurate summaries that communicate the key content, highlight inconsistencies or action items and generally distill the most relevant content is incredibly powerful as a timesaver. In addition, LLMs can be very helpful in creating draft responses to emails, especially.
As a second, but related, category, we often suffer from information overload. LLMs can help reduce that problem by summarizing and highlighting relevant information. Many companies are building LLM-based interfaces to their intranet as a much more effective mechanism to unlock the knowledge and information stored there. LLMs can also help in dealing with conflicting information and advice. Finally, especially personalization of the input based on what the LLM ‘knows’ about you can be extremely helpful and support productivity.
As a third area, decision-making typically requires balancing multiple information sources and deciding on the best course of action despite significant uncertainty. People use LLMs for exploring different scenarios and comparing alternatives. As LLMs can capture sufficient context, especially for complex, multi-faceted decisions, it can be very helpful to be supported in clarifying objectives, alternatives and consequences.
The above naturally results in the fourth role that we identified in this stage: AI as your sparring partner. Whether it’s for creative projects or for getting over a writer’s block, simply framing a question to an LLM and starting to brainstorm with it can allow for much more rapid progress on creative projects of any kind.
The final role is, of course, task execution. LLMs can be quite effective in either assisting or fully automating certain tasks. Whether it’s generating code, filling out forms or reports or even generating templates, they can automate or at least assist in a wide variety of tasks and improve productivity for individuals dramatically.
Of course, as nothing is ever as good as it may seem from the outside, there are things to be aware of. Due to hallucination (which is becoming less prevalent, to be honest), we always need to check and verify the LLM output. Especially when it concerns facts and logical reasoning steps.
Second, it’s important to understand the limitations of models. Especially creative, new thinking and creating things that don’t exist yet is where models are limited and humans are much better, in my experience. Like any tool, LLMs are also to be used where they’re helpful and not elsewhere.
Third, there’s such a thing as “prompt fatigue.” To get good outcomes, you need to work with the prompts that you give the models. If you’re not getting the results you’re looking for, then it’s quite likely that you need better prompts. The good news is that you can use the LLM to structure and create prompts for you.
Although playtime is only the first stage in the maturity model and the implications for the company as a whole might be limited, the impact on individual professionals can be quite profound. Using AI models for automating routine cognitive tasks, managing information overload, improved decision-making, sparring partner or task execution assistant can provide enormous productivity benefits. The main challenge is to use the model more as a thought partner and less as a tool. To end with a quote from Sundar Pinchai: “AI will be part of our future. It’s inevitable.” As the future is where we spend the rest of our lives, we better make sure we can make the best use of this new partner.
