Sponsored We’ve spent the last few years building collaboration systems and filling them with our enterprise content. Isn’t it time that we did something with those documents and emails rather than simply sharing them on message boards and editing them together?
AI could help us use our business information to automate tasks and even prevent employees from breaking industry regulations, if we can only overcome the challenges that stand in its way.
Even though employees love digital collaboration, they don’t always get the tools they’re after. According to Deloitte, 32 per cent of employees aged 40-49 used collaboration tools not supplied by their employer because their own applications had a wider range of functionality than those officially approved at work. This rose to 42 per cent of workers under 30, showing the high demand for more functional collaboration tools among younger employees.
Better collaboration does the same thing as every other useful technology: it gets workers home early by making every day processes smoother. Today, a lot of work still involves human interaction. When Bob sends a change requisition letter to Jan, she signs it and forwards it to Brian, who includes it in the agenda for a weekly meeting. When Bill gets an invoice, he copies key information from it into an order processing system. Wouldn’t it be nice if the collaboration system could handle some of those things for us?
AI – specifically machine learning – is good at analysing mountains of data, and recognising everything from concepts through to images and behavioural patterns. After AI has worked its magic, workflow and process automation can automate key tasks. These techniques combined would seem to be the perfect choice to make processes more intelligent.
The C-suite seems to agree and is having a love affair with the kinds of capabilities machine learning can offer. A 2016 ServiceNow survey of 500 CIOs found 52 per cent advancing beyond automating routine tasks toward making more complex automated decisions. Three years later, Gartner’s 2019 CIO survey found that AI use in the enterprise had grown 270 per cent.
Collaboration suites seem like the perfect place for AI to flourish because they store the data it needs. With so many files, emails, images and audio stored within them, the most valuable collaboration systems should serve as a mine for ML to exploit looking for opportunities to automate.
If companies can bring AI to bear by mining data in collaboration systems, they could deliver several benefits. They could save time by speeding up repetitive workflows, cutting valuable days or even weeks out of everyday tasks. Those time savings could also introduce cost savings. If Bill’s speedier invoice processing lets the company pay a supplier earlier, it could yield early payment discounts.
AI in collaboration and content sharing systems can also make the lives of compliance officers and security pros easier by introducing more data governance and protection mechanisms. If AI can spot suspicious patterns of behaviour in employee interactions and the database store, then it could help spot and avert compliance problems, saving companies from regulator fines and public embarrassment.
This would solve a growing business need. According to the 2019 Datis State of Workforce Management survey, 73 per cent felt prepared to address regulatory compliance this year, meaning that they will be looking for solutions to help them do this at scale.
This sounds great in theory, but what does it look like in practice? Let’s take that compliance scenario first.
Data-loss-prevention systems have detected sensitive information for years using pattern-matching techniques, but a machine-learning system could take things a step further. It might understand the context of sensitive information in a document, message, or email. It could also understand the role of the people who shared it and received it, and infer the implications of that sharing. That could prompt an alert to the employees involved and to the compliance department.
An AI system could also apply storage and sharing policies based on the content of a document far more quickly and at greater volumes than humans could do, with fewer errors.
AI could go beyond simply recognising content in a document by actually validating it. Let’s take Bill’s invoice problem. An AI could extract the information that he would normally copy, and check that it was correct before inserting it into the order processing system. It could learn about historical business workflows surrounding that data and trigger them, perhaps messaging the person responsible for approving the invoice so she could quickly do it from her smartphone.
Hello, is it me you’re looking for?
AI will also help employees find the information they need in unstructured documents that don’t have the explicit fields you might find in an invoice. We can search documents in collaborative content systems today using keywords but try finding all the files sent to both you and Jane in the last month that concerned next quarter’s sales event in Hawaii, but which didn’t discuss budgets.
Natural language processing makes it easier for humans to create that query, and for enterprise collaboration systems to understand it. AI can also find it more easily using a combination of metadata and conceptual understanding based on the information it sees in those files.
For this to work, AI needs to classify and label files, messages, and other records, connecting them with each other to surface relevant information when asked. That’s how a manager will find all documents relating to a project that don’t explicitly mention it. This can go further than text-based documents by bringing in image tagging and speech recognition documents that could mine audio recordings for topics and sentiment. Imagine asking how happy a customer was with the company’s customer service and getting back a sentiment score based on their last few contact centre sessions and emails.