I love productivity and efficiency. I preach it, I evangelize on its behalf. I campaign for its adoption across every enterprise and initiative that seeks my advice and counsel. There is a line between Utility and Assistance, however, which cannot yet be crossed – no matter how many startups try valiantly to ignore the prevailing reality. Utility is a largely passive operation, which must be activated and managed by the user to fulfill its potential. It’s a useful tool such as OneNote or Wunderlist. Assistance is an active function that manifests itself independently, and must anticipate and manage multifarious unqualified scenarios to be truly effective.

The list of Virtual Assistant startups grows daily. It’s the present fad. For every variation that promises to reinvent the VA space yet flames out (Zirtual), another two replace it with air-dancing artificial plums (e.g.: Genee, x.ai). The new holy grail of tech startups is AI virtual assistant apps. For the next 6 months or so, all the early adopters will fall over each other, just to be able to claim they had “Amy”, “Genee”, “Cortana”, “Siri 2.0”, et al, before everyone else. What you won’t hear much about is the fact that all these AI solutions fall far short of useful. Virtual assistants have existed for years, and work with varying degrees of success. Productivity apps have been around for a long time as well, exhibiting capricious achievement in their own right (yet but few pretentions to actually *replace* staff). Zirtual did not fail to provide the services they promised to clients. The company failed because, like so many startups today, it was encouraged to grow too fast, in an unsustainable quest for lightning ROI. The likely result was an inability to meet financial covenants, founders and investors working at cross-purposes, and lack of transparency between stakeholders seeking markedly different objectives. Whoever takes over the operations, such as they are after this negative brand impact, will assuredly restructure for more realistic growth metrics, if any future is to be realized for the employees and their clients.

I have no doubt that after various highly overvalued iterations churn through talented developers, employees, and investors, the chasm between AI VA concept and reality will begin to narrow, such that solutions that provide useful value finally establish themselves on semi-solid footing, and scale sustainably. Until then, you will have to contend with one offering that has access to your calendar, but not the other party’s calendar; another unable to process plain language text or speech; and probably none that take “drive time”, “weather”, or “distance” in to consideration when booking meetings back-to-back, not to mention the probability of client A being notoriously late, or client B correspondingly early, by habit. In short, none will be able to do what a proficient human assistant can do.

AI

Sometime in the future, our human administrators may well be replaced by competent digital, or even robotic, facsimiles. However, the truest measure of a great assistant is their ability to adapt to and accommodate the unexpected scenarios, and no algorithm can proactively absorb this aspect of the job, yet. Artificial Intelligence learns and improves with use, but most companies and executives who require assistants cannot afford to patiently wade through failure, in an iterative quest for efficiency and reliability. If the day comes that Artificial Intelligence Apps crowdsource their refinements, machine learning will accelerate exponentially, and I’m frankly not sure how comfortable I’d be with an employee who mathematically assures me they know what’s best for me, simply because they know more than me.