Chatbots are really just a way to push and pull messages on messaging channels, nothing more, nothing less. How much automation you put behind it is really up t

Chatbots are really just a way to push and pull messages on messaging channels, nothing more, nothing less. How much automation you put behind it is really up to you. And yes, many early examples forgot about their users. Yes, it is cool to write a bot that can do something cool for a fraction of use cases, but as always, solving a real problem trumps being cool. Now chatbots can be used as a content distribution channel (see publishers’ bots), app replacement (Swell is a great example, and they are in the next YC batch, congrats to them), but I’d like to point out how makers miss the point on arguably the largest use case for messaging in the business world today, customer service. Yes, most bots that address this domain can’t live up to the expectation either. And really, just think Job to be Done, the customer wants a quick, accurate and preferably personal reply to their question or issue, it is that simple. So what we (Bicycle AI) did to make our clients better at this Job is we are fusing human and AI for a faster, affordable and scalable solution for SMEs. I hear you say, wait, isn’t this like Digital Genius, well, nope, we integrate with Intercom for example and we have a team of AI supervisors who approve the answers suggested by our AI. This means our clients can provide blazing fast customer service for their customers 24/7 and according to their volumes. Simple as that. We use past conversation history and product docs to generate our suggested answers. We can go live pretty fast and can take care of around 60-70% of incoming conversations, escalating the complex cases after engagement. Aaand, we are in alfa stage with emails. So when it comes to harnessing automation in customer service, we believe that customers should not be the ones dealing with the rough edges.