AI agents – part of everyday life at Coor

AI agents support Coor’s operations in several ways, including automating workflows and reducing risks. They can perform certain tasks faster and with greater accuracy than our employees, who in turn gain more time for less repetitive and more value adding work.

AI-generated illustration of Coor's AI agents Adde and Geoffrey standing in front of an office building and waving

Automation saved almost 10,000 working hours

Coor currently has two AI agents in operation and several more under development. Together with our other robotic process automation (RPA) solutions, the automations completed over 70,000 tasks in 2025, saving almost 10,000 working hours. This year, time savings are expected to increase substantially as existing agents scales in functionality and volume along with additional agents being put in production.

Thobias Björk

– Automation is nothing new for us at Coor; it’s something we’ve been actively working with for many years. What’s new is that, with the help of AI, we can build reasoning into our automations, which opens entirely new possibilities and use cases, explains Thobias Björk, Head of Data Analytics & Integrations at Coor.

AI is transforming the way work is done at Coor

A common feature of Coor’s AI agents is that they are part of the regular workforce. Each one has a unique name, a responsible manager, and specific tasks. The agent Adde automates administrative workflows related to assignments and invoicing. By generating work tasks and ensuring correct compensation, processes become more consistent and quality-assured. The agent Geoffrey handles incoming service requests by interpreting content, identifying the correct asset, and registering cases. This enables faster handling and reduces manual work.

– AI does not replace our work – it changes how we do it. By letting AI handle recurring tasks, we free up time and capacity among our employees, allowing them to use their expertise for more challenging tasks that create real value for our customers and investors, says Thobias Björk.

Coor’s customers receive faster support

Geoffrey is currently the agent with the greatest potential to scale and create value beyond Coor. He is already being used by several customers in Sweden and will soon be launched in selected contracts in Norway and Denmark.

– The agents’ capabilities are continuously developing, and Geoffrey is well on his way to becoming an expert in certain tasks. For example, Geoffrey flags cases in the queue much faster than a human, ensuring that high-priority issues are picked up earlier, which is creating tangible value for our customers, explains Thobias.

The next step is to expand Geoffrey’s footprint by implementing the new capabilities. He can then be configured to the specific tasks that each new customer deployment requires.

How do I start developing AI agents?

  • Start with the process so that you don’t spend time optimising something that isn’t good to begin with.
  • Data quality and order is as important as you have read about, the agents will not be successful without proper information.
  • Don’t hesitate to stop development when costs escalate or results fall short of expectations.
  • Use older LLMs for simpler tasks to keep costs down.
  • Think big, but start small – even minor improvements can make a big difference in everyday work.

 

Coor's AI agent Geoffrey