At Cognel, we spend a lot of our time on one question: how do we help business users make better decisions? Below that sits a more fundamental one: what is a good decision in the first place?
Decision science gives a clear starting point. Ronald Howard, who coined the term "decision analysis" in the 1960s, drew a firm line between the decision and the outcome. You can decide well and still get unlucky. What you control is the process. Research since then adds two requirements: the process has to fit the problem at hand, and it has to bring in several points of view before it settles on one.
Many decisions meet neither. Paul Nutt (1999) studied more than 400 decisions by top managers: about half were not fully in use two years later, and only 29% of the decision makers had considered more than one alternative. AI makes this worse, because it makes it easy to skip the process entirely and take the first answer.
For us, that translates into two design questions.
(1) Which Guides do we need to offer? That depends on who we build Cognel for, so it is really a targeting question.
(2) How do we build a Guide so that users stay engaged with their problem and end up with a solution they own?
The second question is the hard one. Two things matter here: how easy a tool is to use, and how much the user owns the solution at the end. Today, you mostly get one or the other:
Chatbots are extremely easy to use, but the solution belongs to the machine. You get an answer, and you don't own it. And a solution you did not build is a solution you will not defend in the next meeting.
Cognel today sits in the opposite corner. Users work through their problem and build their own solution, which stays theirs. The price is effort: it is not always easy to use.
The corner where a solution is both easy to reach and yours is still largely empty. Our Guide design principles are our attempt to get there. Whether that corner is reachable, and whether it is even desirable, is what we are exploring with Prof. Eva Bittner and Dr. Lucas Memmert.
#DecisionMaking #Reasoning #HumanAICollaboration #Cognel