This article is more than 1 year old

If only something could intelligently analyze our mountains of paperwork. Hey, let's ask our new friend, machine learning

Compliance, sharing, collaboration, process optimizing... there's so much AI can do in modern enterprises

Maybe it’s OK for AI to read your emails after all

AI could mine those unstructured documents for workflow-related content, or at least alert employees that something in the document might be relevant. In a supply chain logistics context, if a machine-learning-driven bot saw a file from a certain group containing information that suggested a delivery delay, it might alert the operations manager and the procurement group, suggesting another supplier who could deliver on deadline.

An algorithm might help field service engineers by mining emails and phone calls between the customer and the customer service team for information about the object being serviced and the customer’s experience. A workflow like that could help raise single-visit success rates.

In an ideal world, machine-learning algorithms would go further still, giving us information and suggesting actions that we didn’t know we needed. Machine Learning excels at analysing historical data to find baseline activity. If you have a meeting every Monday, it’ll pick that up. It’s a short step to having it collect the information for that meeting using textual analysis. If you’re sick, it’s not unrealistic for the AI to examine everyone’s schedules and rearrange the meeting for you.

Joining the dots

This is all very exciting, but many companies can barely make their email and calendar systems work together. IT systems seem almost hard-wired to resist collaboration. Developed organically over the years, companies’ existing document management systems are often information silos.

Things aren’t much better in the new, cloud-based era. The Deloitte survey already showed us that employees use collaboration systems not officially mandated by the company, making it hard to standardise on any one platform. The Datis report found 30 per cent of people cite too many internal systems or information silos as the greatest obstacle to cross-departmental collaboration.

Integration with back-end transactional applications is even more challenging. With so many bespoke and off-the-shelf operation systems in play, what chance do companies have of pulling all this stuff together?

There are some potential fixes. Single-vendor collaboration hubs are becoming more functional, enabling companies to give their employees more capabilities on a single system and preventing the rise of alternatives.

Then, there are bilateral integrations between different collaboration companies. Slack has made a point of connecting with dozens of third-party cloud services, making it easier to get information in and out of the system.

Finally, there are cloud-based integration platforms that focus on tying together various applications with AI services. Large tech players are exploiting their partnerships in AI and their strong cloud presence to bring AI capabilities to a broad range of collaboration and content management applications. Box, leveraging technology from Google Cloud, Microsoft Azure and IBM Watson in their Box Skills framework, is a case in point.

Enterprise collaboration and AI are still new bedfellows, and AI is still finding its feet in the enterprise. Over time, companies may find value in combining the two, creating new approaches to workflow automation and compliance that reduce human error and speed up the business.

The closer the integration, the more skill it will take. Understanding sector-specific concepts is important when trawling, say, healthcare or financial services documents for relevant information. This means getting data scientists to work with domain experts. It means continually refining the Machine Learning algorithms to make sure it finds the right elements among the thousands of documents available and does the right things with them.

Combining AI with collaboration and enterprise content is a slow journey and requires heavy investment. But perhaps processing Bill’s invoice automatically and paying a favourite supplier before lunch is a small but good first step.

Sponsored by Box.

More about

TIP US OFF

Send us news