Title: Gazing into the Crystal Ball: 2025 Projections for Autonomous Artificial Intelligence
Shailesh Manjrekar is the Chief Marketing Officer at CloudFabrix, a pioneer in Robotic Data Automation Fabric and a leading voice in AIOps. As the world continues to evolve, with wars brewing and significant technological transformations, we're here to provide some clarity on what to expect in 2025.
2025: The Year of Multi-Agents
At its core, an agent possesses the ability to ReAct (reason and act), breaking down tasks and executing them independently. This form of AI operates on various levels of automation, range from Level 1 to Level 5, based on the human involvement required (human-in-the-loop). Agents use LLMs (large language models) or SLMs (small language models) for decision-making, making automation contextual.
Multi-agents are becoming a necessity as software services providers. Several digital workers will be collaborating and coordinating to deliver optimal outcomes. In a previous article, I discussed agentic AI and its importance.
Agentic AI to Transform Vertical Business Models
Experts project that AI will disrupt vertical industries with a predicted $4.6 trillion opportunity. Instead of focusing on seats or platforms, licensing will be tied to outcomes. Fintech and telecommunications industries are among the early adopters, with each $1 investment yielding 3.7x return.
The Gap Between AI Co-Pilots and Agentic AI
While AI co-pilots and assistants primarily focus on conversational queries and human-in-the-loop recommendations, agents analyze, judge, and then autonomously act. By breaking down conversational queries into thought chains (COTS) or thought graphs (GOTS), agents can execute tasks more effectively before replying. LLMs are used for reasoning, while SLMs (small language models) or LAMs (large action models) are responsible for task execution.
The Emergence of the Agentic AI Stack
The agentic AI stack can be divided into three main categories: data fabric for AI, agentic AI workflows, and agentic applications and workflows.
- Data fabric for AI: Data source integration, data enrichment, and data management are crucial parts of contextualizing data for agents, with data fabric architectures and vectorDBs leading the way.
- Agentic AI workflows: Emerging AI frameworks and models, like OpenAI O1, will increasingly be used in conjunction with orchestrators, providing a platform for artificial general intelligence (AGI) or system-level thinking.
- Agentic applications and workflows: AI-infused applications will become the norm, with domain-specific vendors leading the charge in developing agentic workflows and applications.
Movers and Shakers
Major application vendors, such as Microsoft and Salesforce, have already embraced the use of agents, leading to the creation of enterprise applications and intra-marketplace agents.
To close, leaders should define business objectives and KPIs for agentic workflows, establish clear operating models, and invest in responsible AI, privacy, and security practices. Get in touch with our exclusive community for technology executives, the Website Technology Council, to join the conversation. Do I qualify?
Shailesh Manjrekar, as the Chief Marketing Officer at CloudFabrix, might discuss the role of agentic AI in transforming various business models in a future marketing campaign. In a presentation, Shailesh Manjrekar could delve into how agents and multi-agents are becoming essential in delivering optimal outcomes for software services providers.