About a year ago, Facebook launched a platform supporting chatbots, sending ripples of excitement around a technology some claimed had the potential to replace apps.

However, the launch of the platform was “over hyped very quickly,” the company’s VP of messaging products David Marcus himself admitted in December.

The social media giant had not been the first to talk about bots (although the hype did reach fever pitch after its announcement). While messaging apps Kik and Telegram had begun embracing the technology, Microsoft was also talking up their potential.

In fact, it did so not long after controversy surrounding its Tay chatbot, which had been responsible for what the company referred to as “unintended offensive and hurtful tweets” soon after it launched.

The problems with Tay were a major indication the technology behind bots, such as machine learning and artificial intelligence (AI), had a long way to go.

So will bots really ever catch on? Is it simply a case of waiting for the tech to become advanced enough or is it just the kind of service that will suit only certain, niche verticals?

The answer, it seems, is both.

In February, App Annie CEO Bertrand Schmitt (pictured below) told Mobile World Live there is still a lot of work that needs to be done before bots can become efficient and understand what a user wants.

Arguing against claims bots could replace apps, he said there is room for integration between the two, but “apps are here to stay” because they offer a graphical user interface which is easier to use and places less cognitive load on users.

When writing this blog, I caught up with Schmitt again. He explained much of the discussion around chatbots in 2016 centered on conversational commerce.

However: “The counter-argument to this is the very existence of the on-demand economy. Food delivery and ride sharing apps exist because they remove friction – they replace conversation with a graphical user interface. Instead of talking, you navigate around a screen and it saves man hours on both sides of the exchange.”

So bots are not going to be as pervasive as apps because they simply don’t fit every purpose.

This begs the question: what are the best use cases for bots?

Service industries, for one, should be jumping on the bandwagon as it is predicted chatbots will help sectors including healthcare and banking save over $8 billion per annum by 2022, up from $20 million in 2017.

According to Schmitt, the power of bots lies instead in making existing conversation more efficient – for example, customer and after-sales service.

He also said bots present an opportunity for businesses to reinvent processes that have not yet transitioned to mobile-first, including banking, insurance, healthcare and many more legacy businesses.

In India, finance company Tata Capital’s chatbot can handle more than 70 per cent of routine customer queries including product information, application tracking, sales and after-sales service.

When Appy Pie launched chatbot functionality, its CEO Abhinav Girdhar said: “A small business app powered by a chatbot will improve its efficiency significantly by diverting mundane and easily asked questions to the chatbot. I believe artificial intelligence has the potential to give every worker a virtual assistant.”

On the tech side, bots are not yet sophisticated enough to understand human context, tone and emotion, and are limited to scripted conversations. Even in that situation, writing scripts is no easy task and a relatively new job description people are still figuring out.

What’s more, according to Mitch Lieberman, director of research at G2 Crowd, a business solutions provider, operational silos are keeping bots from delivering the level of customer service companies hoped they would.

He pointed out bots are different from virtual assistants since the former is driven by an intelligent agent and is a more natural, intuitive, friction-free system of engagement, allowing for seamless transition between system interactions and agent.

His advice to businesses is to start small and augment the service people provide, rather than replace them completely from the onset. He also suggests companies should train a small group of agents to be ready to receive a hand-off from a bot.

In order to be successful and avoid inconsistent brand voice or the need for customers to repeat information, companies should pick a straightforward customer support use case, he added.

Schmitt echoed this sentiment, stating bots should augment, rather than try to replace, apps in areas of high friction.

In fact, bots can act like an “automatic triage system”, reducing traffic on queues more efficiently for customer service centres, and even interactions with medical practitioners.

At the end of the day, bots will only attract users if they offer an experience which is easier and more convenient than the existing alternative.

As Schmitt put it, “how is using a weather bot a meaningfully superior experience compared to a weather app?”

However, scheduling a doctor’s appointment or sorting out problems with a credit card are more complicated scenarios and areas where bots can make an impact.

A little over a year after first announcing the bot platform, Facebook is still betting big on bots, launching Discover, a hub inside Messenger for unearthing new chatbots, and taking things up a notch by working on bots capable of negotiating.

It appears the social media giant is also likely to lead the charge in this field, although Schmitt believes use cases will remain limited to sending notifications and routing users to the right human point of contact, at least in 2017.

The editorial views expressed in this article are solely those of the author and will not necessarily reflect the views of the GSMA, its Members or Associate Members.