Don't Build Black Boxes

What we found with Viv (which mirrored what was found with Siri / Alexa / Google Assistant / etc) is that people do very poorly with a black box.

They either project too many capabilities onto it which it's incapable of doing (so they get frustrated and lose trust), or they are paralyzed with not understanding what it can do so they do very little with it at all (missing out on potential wow moments).

Just like with teaching kids, you need to scaffold people from where they are stuck today to where they could be with a little help. Most can't jump over a chasm, so instead you have to build tiny scaffolds and let them slowly build trust and confidence. Remember that most people use Excel to create tables, not for calculations. These table builders are your early users too, so stop forcing them to immediately do calculations.

Chatbots today are more like a black box / chasm, which is why adoption is okay but not nearly as good as it could be if you took more time to build skill affordances into each of your "magic" AI features. Magic is a poor word choice because it implies that it's hard to understand what it's doing and how to succeed at it; people still need configurations and controls to inject their opinions, standards, compliance concerns, etc. and they need others to do some of that work for them (e.g. bootstrap them with examples / templates / community variants / activation onboarding).

People are literally telling you where they're stuck while shouting "I'm stuck!!! Back here, hey! Here I am. Please help!" while too many product builders are shouting back "Okay, just jump on over here to my shiny thing!! Come on, figure it out. I'll be waiting" (on distribution, mainstream adoption, SEO, the customer's pain points to change).

It's up to us to listen and build the scaffolds.

-- Rob