Agencies spend too much time doing manual work.
One of the biggest time sucks? Transferring data files between enterprise systems that don’t talk to each other.
Max Cheprasov, now an exec at the Dentsu Aegis holding company level, recognized these inefficiencies while working at Dentsu agency iProspect starting in 2011. He set out to document and standardize processes while outsourcing inefficient tasks so that employees could focus more on strategic client work.
Eventually, he brought artificial intelligence into the agency’s workflows, specifically natural language processing and machine learning, which helped accelerate the ability to interpret data, derive insights and generate reports.
By 2017, automation made iProspect the most profitable agency within Dentsu and Cheprasov was promoted to chief automation officer in order to scale his vision across the network. He drafted an eight-year plan, through 2025, with the ultimate goal of integrating AI wherever possible.
“The opportunities are limitless,” he said. “AI and automation should be injected in every process and workflow.”
By automating mundane tasks, AI helps agencies deliver work and insights to their clients faster.
When filling out RFPs, for example, teams often spend weeks on 50-page documents that are chock full of standard questions. But by partnering with workflow automation platform Catalytic, Cheprasov’s team employed AI to fill out standard information on every RFP automatically. Subject matter experts then look over the answers and tweak them where necessary.
That process condensed the time it takes to fill out an RFP from weeks to several minutes, Cheprasov said.
Dentsu also uses Catalytic to automate campaign reporting so that agencies can deliver insights to clients quicker and more frequently. The platform automates tedious work, such as transferring and validating data files and uploading them into billing systems, thereby reducing manual effort by between 65% and 95%.
“Data collection, processing and reformatting should be automated, because it’s a horrible use of people’s time,” said Sean Chou, CEO of Catalytic.
In late 2017, Dentsu first began rolling out its strategy in Japan, where it identified 900 processes that were ripe for automation. The system is now also in place in the United States and Brazil, and markets across Europe and other parts of Asia are starting to get involved.
Today, Dentsu is exploring how to use AI to build automated processes for agency workflows that haven’t been documented before. Using computer vision and natural language processing, Cheprasov’s team can analyze keystrokes to create process maps that it can later automate.
“It’s a good baseline for what people do, how they do it and how it should be redesigned,” he said.
Dentsu’s long-term goal is to arm all of its employees with a virtual assistant, accessible through a conversational interface, that can carry out manual tasks and tap into a central “brain,” where all of the agency’s processes live. To do that, Dentsu will train staff to use low-code or no-code systems, so they can engineer assistants and document new processes on their own.
This could help automate between 30% and 60% of what Dentsu employees currently spend their time on.
Stats like that can be scary for agency employees, but Cheprasov’s goal is not to do away with jobs.
Mind-numbing tasks are generally spread across roles, rather than comprising a single person’s entire job, and a lot of this grunt work has already been sent offshore in any case.
“The mission is to elevate human potential,” Cheprasov said, “not to eliminate it.”