May 18, 2017 | Written by: Simon Rabone
Categorized: Automation | C-Suite
Share this post:
Despite the attention-grabbing headlines, suggesting the workforce may one day be replaced by an army of robots, the reality is more complex. Fewer than 5% of jobs are fully automatable, according to McKinsey – but most roles have the potential to automate a significant proportion of their activities.
This is the case in IT, where many repetitive tasks can be at least partially automated. Which means people need to work alongside automation, incorporating the activities executed by bots into their workflow. Adopting new technology is challenging though, especially if it appears to be undermining our own usefulness. How can we ensure our people adapt their ways of working and make the most of these new capabilities?
At IBM Global Business Services, we’ve deployed automation into hundreds of the accounts we support. Those include process automation, where we’re using tools, from partners at BluePrism, IPCenter and Automation Anywhere, to execute repetitive tasks including system monitoring or helpdesk ticket triage. And our IBM Automation with Watson capabilities include Agent Assist, which guides Helpdesk advisors to resolve tickets quickly and consistently, and Coding Assist, helping junior programmers tackle more complex coding work by providing insights and knowledge.
All of these use cases and many more mean we are constantly challenging our teams to change the way they work. To get the full business benefit from our significant investment in automation, we need to win the hearts and minds of our people.
Here are our top five keys to successful automation adoption:
Take the time to educate the team on why a new tool is being introduced. One Delivery Leader said ‘We told the team that using Cognitive is the way IBM is going, this is the way IT is going, and so we should take the opportunity to lead the way and be a flagship team’. Other teams have re-routed the hours saved on repetitive tasks into enhancement work, or reducing the backlog of maintenance tickets, freeing them up to be more proactive. Sharing those impacts upfront help motivate, answering the question ‘what’s in it for me?’.
2) Start small and grow
Rather than rolling Cognitive Agent Assist out across one large account all at once, they worked with one functional support team at a time. This allows some ‘hot housing’ (see 3, 4 and 5 below) to bring that team up to speed before moving onto the next functional area. Other teams have worked in an agile way, with multiple sprints ensuring a high quality series of smaller deployments of new Robotic Process Automation tools, using a ‘deploy, learn, revise, deploy’ approach.
3) Set a new standard
Use of Coding or Agent Assist should be introduced as a standard operating procedure for handling all incidents and small routine change requests. Driving usage is critical to improving the quality of the tool, as it increases the number of opportunities Cognitive has to receive feedback and to learn from the answers it is providing. Daily peer reviews of ticket resolution details act as a control point, so pockets of low usage can be investigated.
4) Showcase successes
As well as looking into low usage, it’s equally important to highlight high usage. Team members who are using Coding or Agent Assist extensively and entering high-quality information into their ticket resolution details are held up as a success for others to learn from.
5) Continuously improve
Cognitive Agent Assist has continuous improvement built in – every time it provides an answer, it gets feedback on the answer which will increase the confidence level in that answer next time the same question is asked. It also learns from the corpus, or body of knowledge, which ingests ticket resolution details. So it’s important that the team not only ask Cognitive questions but also provide feedback on the answers given and enter good quality information about how their incident was resolved into the ticketing tool. Some teams begin with manual uploads of their ticketing data, to allow for a peer review. Once they are satisfied the standard is high, this feed can be automated, providing a seamless feedback and learning loop between Agents and Cognitive Agent Assist.
Are you deploying automation tools in your workplace? What has worked for you?