2024 is seen as a breakthrough year for artificial intelligence and agent use case experimentation, while 2025 is predicted by industry experts to be the "Year of Agents," a year when various AI pilot projects and experimental results will converge and start to show a return on investment. As more companies apply artificial intelligence to enhance productivity and optimize workflows, AI agents will become central to driving innovation and efficiency.
Image Source Note: Image generated by AI, licensed from service provider Midjourney
More Deployment and Productivity Drive
AWS Vice President of AI and Data Swami Sivasubramanian pointed out that 2025 will be a crucial year for productivity, with executives increasingly focused on the cost-effectiveness of AI. He stated that as multiple agents are deployed in corporate workflows, the focus will shift to enhancing the efficiency, accuracy, and productivity of these agents. "In the world of agents, workflows will be reimagined, and businesses will begin to focus on achieving a fivefold increase in productivity."
Akshay Krishnaswamy, Chief Architect at Palantir, also noted that decision-makers are showing strong interest in the actual impact of AI investments, especially those senior executives who are not directly involved in technical decisions. "The enthusiasm for experimentation has gradually waned, and executives are eager to see real returns on investment," he said.
The Rise of Orchestration Frameworks
As we enter 2025, managing AI agents and applications will face greater challenges. With growing demand, companies will have an urgent need for efficient orchestration platforms. Chris Jangareddy, Managing Director at Deloitte, stated that AI companies like LangChain will face an increasing number of competitors, with new orchestration tools continuously emerging. "Many tools currently on the market are trying to catch up with LangChain, and more new players will appear in the future," he explained.
Although LangChain is currently the most popular tool, some companies are still exploring other solutions, such as Microsoft's Magentic and emerging platforms like LlamaIndex. Matt Wood from PwC indicated that orchestration frameworks are still in the experimental stage but will undoubtedly continue to evolve, with more options becoming available.
More Powerful Agents and Integration Capabilities
As AI agents become more deeply integrated into corporate workflows, effectively integrating agents across different systems and platforms will become an important task. Platforms like AWS and Slack have already introduced connection tools with Salesforce Agentforce or ServiceNow agents, enabling companies to easily transfer contextual data between different platforms.
However, as the complexity of agent workflows increases, supporting these integrations and ensuring smooth coordination of agents handling multiple platforms will be a focus of technological development. Newly released reasoning models, such as OpenAI's GPT-3 and Google's Gemini 2.0, will make these agents smarter and more efficient.
Despite ongoing technological advancements, Don Vu, Chief Data and Analytics Officer at New York Life, cautioned that if companies cannot ensure their employees effectively use AI tools, all efforts will be in vain. "Changing human behavior is more challenging than deploying applications, and the 'last mile' challenge will remain a persistent issue in 2025."