Why do chatbots work? Because a quick chat resolves a lot of problems.

At the workplace, we spend a lot of our time managing processes and slogging through impersonal workflows to complete mundane but essential tasks. It would be better to handle these with a chat, and that’s exactly what conversational workflow promises.

Chatbots’ capabilities have expanded significantly in recent years. Last year’s chatbot announcements by Facebook propelled the technology further up the hype curve. At the peak of that hype, we saw customer-focused chatbots that did little more than prove that the technology existed and worked. Too often, they were solutions looking for a problem. Now we are moving beyond the hype and creating bots that are genuinely useful in either enterprise workflow or customer engagement. The technology is now solving real problems.

With such technology in the workplace, you might ask a chatbot to give you company revenue for the last quarter. The CFO and the CMO might both ask for that information, but what they would want to know next would probably differ. A smarter bot, one that understands your role in the business, might suggest the next bit of information: “Would you like to see that broken down by division?” for example.

The task of onboarding a new employee can be made much simpler by a chatbot that asks for the name of the new employee and the department they are joining and then automatically adds them to the relevant groups and business systems, based on predefined rules. Likewise, employees can request vacation time easily through chat, with a bot asking for the relevant dates, checking that the employee has sufficient vacation time remaining, and then passing the request to a manager for approval.

The benefits are obvious. A conversational workflow lowers the cognitive load on your employees, makes them more productive, and helps them get things done in a natural way — through conversation. It doesn’t require businesses to reinvent their existing workflows. Instead, it enhances existing ones.

In our personal lives, we are already very comfortable with chat interfaces. Worldwide, 1.4 billion people are using messaging apps. These apps are bigger than social networks, hence Facebook’s expansion into Messenger and its acquisition of WhatsApp. And we aren’t just talking to friends and family. A third of us would now rather talk to a retailer via chat.

As chatbots acquire more AI smarts, they are able to offer more intelligent conversations. Using natural language processing, they can recognize a question without requiring the user to be precise. It is increasingly easier to connect these bots to databases, systems, and services that give them the knowledge they need to perform useful functions.

Chatbots can also draw on relevant information that they have already been given. If you started the interaction by giving your name, the bot will know that it doesn’t need to ask your name later on. A chatbot that detects that you are using a Swedish keyboard might offer to answer in Swedish.

Companies can set rules for how chatbots respond to different customers. For example, two customers contacting a company to complain about a missing delivery might get different responses based on their order history. A casual customer might just get an apology from the chatbot and a promise to ship a new order right away. A regular customer could be offered a voucher and guaranteed next-day delivery of the new order.

One tricky thing about workplace chatbots is that company processes can differ widely and often need to tie together multiple tools in unique combinations. That means they often need considerable customization before they fit the intended task. That’s why partnership and customization is so important.

What we’ve discussed so far is conversations triggered by the end-user. The future of chatbots is proactive. If your flight is delayed, a chatbot could contact you and perhaps offer to re-book you on a later flight. If the deadline for your expense claim is due and you haven’t filed it yet, the chatbot — which has access to your emails and calendar — might gather some basic information and remind you to fill in the rest.

That stereotypical request a boss makes of the assistant (“Cancel my appointments for the rest of the day!”) will be easily handled by a chatbot, which will not only contact your jilted colleagues but also check everyone’s calendar so that it can reschedule.

You will still have to do your work yourself, however. Some things don’t ever change.

Tony Lucas is the cofounder and CEO of Converse.ai, a chatbot development company.

Above: The Machine Intelligence Landscape This article is part of our Artificial Intelligence series. You can download a high-resolution version of the landscape featuring 288 companies by clicking the image.