Automate Tasks with AI Agents, Not Entire Jobs
Automating tasks with AI agents is revolutionizing industries, yet many companies face challenges in realizing their investments. A notable case is Klarna, a finance company that learned the hard way that attempting to replace entire jobs with automation often leads to rehiring. After aggressive automation efforts, Klarna reverted to bringing back human employees, indicating a broader trend among businesses navigating the use of AI agents.
Lessons from AI Implementations
According to McKinsey’s analysis of over 50 AI agent implementations, the key to success lies in how companies approach automation. Those that automate specific tasks within restructured workflows tend to thrive, while others aiming for total job replacement often struggle. This distinction is crucial in leveraging AI technologies effectively.
The Shift Towards Task Automation
The landscape of AI integration is evolving rapidly. Henrik Landgren, Co-founder and CPTO of the Swedish finance company Gilion, emphasizes that strategies adopted just six months ago may already be outdated. Companies must adapt their conversations around AI capabilities, focusing on agentic AI instead of solely generative AI or traditional machine learning.
Innovative Use of AI Agents at Gilion
Gilion is a prime example of a company that has successfully integrated AI into its operational framework. By using data to aid investors and lenders, Gilion collects various types of information, including:
- Payment data
- Marketing data
- Accounting data
- Product usage data
Using machine learning, Gilion can predict outcomes with 90% accuracy over 12 months. However, the real breakthrough came with the integration of generative and agentic AI, greatly enhancing product experience and improving customer value.
Task Management with the MECE Framework
The MECE (mutually exclusive, collectively exhaustive) framework guides Gilion in managing its 82 AI agents effectively. This structure allows each agent to tackle a specific aspect of the investment analysis process, making sure that all tasks contribute to the final goal without overlap. The process involves:
- AI agents performing designated tasks
- Agents reporting findings to managers
- Condensing outputs into interactive investment memos
This structured approach ensures that investment analysis is thorough and customizable, unlike the challenges posed by attempts to automate entire roles. It combines automation with human oversight, fostering collaboration between AI and human experts.
A Data-Driven Investment Strategy
Landgren views this data-driven methodology as vital for making informed investment decisions. By minimizing human biases, it boosts the likelihood of identifying worthwhile opportunities while reducing the risk associated with poor investments. Even personal biases can be challenged, as Landgren noted when reassessing emotionally appealing investments against analytical data.
In conclusion, the future of automation lies in enhancing human capabilities with AI agents instead of attempting to replace jobs entirely. For established companies, embracing task-based automation without resorting to staff cuts is essential. Understanding AI agents as collaborators rather than replacements is crucial for successful implementation and long-term sustainability.
