This Chatbot Developer is Disrupting the R&D Process
Innovation in 2018 looks a lot different than it did a few decades ago.
It used to be that research and development happened in large organizations under loose governance focused on big ideas. Today, R&D is done in little, well-financed startup labs driven by demands for immediate revenue. Those are two opposing strategies.
The result of the change is the proliferation of small ideas. This new, transactional approach may produce tremendous innovation; however, the question is whether focusing on immediate returns is good for anyone besides investors. The jury is still out.
In a recent talk at the Santa Fe Institute, the preeminent academic-research center exploring the frontiers of complex-systems science, Alan Kay outlined four components of long-term R&D: visions, rather than goals; people, rather than projects; milestones, rather than deadlines; and funding future researchers. Kay is a pioneer of computing innovation who was describing how the hyper-productive research labs at Xerox PARC worked.
In HRExaminer’s 2019 Index of Intelligent Technology in HR, I outlined the differences between legacy players and small, well-heeled startups. The former has a strategy that resembles the long-range approach, while the small players are focused on immediate revenue. The big guys have data and sometimes, vision. The little players are scrappier and more tactical.
Kronos, Ultimate Software, ADP and Workday are fantastic examples of companies that pursue vision in their development processes. Each uses its core values and experience to create a differentiated view of the future and what will be most important, while their R&D programs are rooted in the organization’s core values.
It is extremely unusual to encounter a smaller company, focused on building intelligent tools, that has a big idea. Vision is usually underemphasized, while tactical success is celebrated. That poses a problem with technologies as primitive as AI is today.
There are some exceptions, particularly with chatbots.
Chatbot developers face similar pressures as designers of other AI applications. The problems are numerous and the challenges significant. Currently, the best way to guarantee chatbot performance is to limit it to precise decision trees that solve very small problems. The hope is that AI will increase employee access to and utilization of company data, but those remain elusive goals.
Recently, I’ve been talking with David Karandish, the CEO and founder of Jane.ai, which is different from other companies in a number of ways. Based in the Midwest, it is modest in its assertions and expansive in its vision. Karandish sees the AI question as a 20-year development journey that is just beginning.
Karandish is building Jane.ai as a project that discovers and breaks through the limitations encountered by intelligent technology. “Company information is like a tangled bowl of spaghetti,” he told me. “There is a huge difference between building a chatbot and building an enterprise-grade knowledge-management system.”
By focusing on industries that change slowly (such as education, financial services and utilities), he is focusing Jane’s evolution on markets with stable structures in their information. In a nutshell, Jane.ai provides intelligent access to company HR policies in whichever digital workspace the company chooses.