In a bold move that could reshape the future of work in the financial sector, banking giant Citigroup has announced the launch of a pilot program involving 5,000 employees to test “AI agents,” an advanced technology capable of independently executing complex tasks, marking a leap from simple chatbots to intelligent digital assistants.
Article contents:
- Introduction: Beyond chatbots
- What are “AI agents”?
- The pilot program at Citi: A step toward the future
- How the new technology works
- Cost and ROI challenges
- Impact on the workforce: Will AI replace humans?
- Conclusion: Citigroup bets on the future of autonomous work
- Frequently asked questions
Introduction: Beyond chatbots
While most companies are still exploring the uses of basic chatbots, banking group Citi is moving toward a more ambitious transformative technology that has so far proven elusive in many workplaces. Companies are betting that one day, AI agents will be able to complete complex tasks completely independently, just like human employees, by chaining a series of separate tasks together and accessing multiple internal company systems.
What are “AI agents”?
An AI agent is more than just a chatbot. It is an intelligent system that can understand a complex goal, break it down into smaller steps, and then execute those steps independently. For example, instead of asking it to answer a single question, you can ask it to “prepare a complete profile on client X,” and the agent will search public and internal databases, aggregate the information, analyze it, present it in a comprehensive report, and perhaps even translate it into another language, all from a single prompt.
The pilot program at Citi: A step toward the future
Citi announced it is taking steps to make this reality happen. Beginning this month, it will start testing new “agentic” capabilities within its AI platform, which it has been developing over the past two years. With the new update, users will be able to direct the AI tool to complete multiple tasks and access multiple corporate systems with a single prompt, according to David Griffiths, Citi’s chief technology officer. He added: “Two years ago, you could do agentic things with early versions of the models available then. But they weren’t always very reliable… but now they are.”
How the new technology works
Griffiths said Citi’s internal platform uses a range of models, including Google’s Gemini and Anthropic’s Claude. With the new capabilities, a user can direct a tool within the platform, known as “Citi Stylus Workspaces,” to search for a specific client, build a profile for them from publicly available data and multiple internal data sets, and translate it into a foreign language, all in a single step. This was possible before the recent update, but it required separate human direction at each stage of the process.
Cost and ROI challenges
One of the biggest challenges is cost. Running agents can become very expensive, very quickly, especially as they handle longer and more complex tasks. Larger tasks require more “tokens,” which are the unit of measurement for the cost of using artificial intelligence. Griffiths said the company has integrated strict cost caps into the systems. However, he added that because the cost of using the models is dropping so rapidly, it is very difficult to come up with up-to-date and accurate ROI calculations.
Impact on the workforce: Will AI replace humans?
This development inevitably raises questions about the future of jobs. If tools become advanced enough to work for hours or more, these efficiencies could begin to influence workforce conversations. Griffiths said: “Does that mean we need fewer people? I don’t know.” He added: “It certainly means we’re going to get a lot of work done. And we’ll see how the workforce evolves with this massive push in capability that we’re getting here.”
Conclusion: Citigroup bets on the future of autonomous work
Citi’s pilot program represents an important moment in corporate adoption of advanced AI. By moving from simple tasks to complex, autonomous operations, the company is exploring the potential for massive productivity gains. The entire financial industry will monitor the results of this experiment closely, as it could set the course for job evolution and efficiency in the banking sector for years to come.
Frequently asked questions
Q1: What is the difference between an AI agent and a chatbot?
A1: A chatbot is designed to answer specific questions or execute simple commands. An AI agent is designed to understand a complex goal, formulate a plan to achieve it, and execute that plan across multiple systems independently.
Q2: What kinds of tasks can AI agents perform in a bank like Citi?
A2: They can perform tasks such as preparing detailed customer reports, analyzing market data, assisting with compliance processes, summarizing complex financial research, and automating large parts of back-office operations.
Q3: Is this technology safe to use in the financial sector?
A3: Security is one of the biggest challenges. Citi is developing this technology on its own private internal platform to ensure sensitive customer data remains secure. This also includes setting strict limitations on what agents can access and do.
