Session brief
The missing pieces of workflow automation — Shirsha Chaudhuri, Thomson Reuters Labs
Overview
This talk addresses the current limitations in achieving comprehensive workflow automation with AI agents. While generative AI, RAG, and agent frameworks have advanced significantly, enterprises are still missing key components to fully reimagine and automate complex business processes. The presentation highlights the gap between existing stable technology stacks, like mainframes, and the potential of agentic workflows, emphasizing the need for better connectors, reliability, standardization, and collaborative user experiences.
Who should watch
- AI Engineers
- Product Managers
- Builders looking to automate enterprise workflows
- Those struggling with integrating AI agents into existing IT systems
- Individuals seeking to understand the challenges and missing pieces in AI workflow automation
Key takeaways
- The journey of AI adoption in enterprises has progressed from democratizing generative AI to implementing RAG, prompt engineering, and now agentic workflows for automating entire processes.
- A significant missing piece is the development of robust connectors that bridge the gap between legacy systems, such as mainframes, and modern AI agentic workflows.
- Reliability and demonstrating clear ROI are critical stumbling blocks for stakeholders when considering the adoption of AI agents for automation.
- There is a need for visionaries and subject matter experts to collaborate in reimagining business processes with AI, moving beyond just automating individual tasks.
- Standardization in how agents are built, packaged, and deployed is crucial for wider adoption and integration within established tech ecosystems.
- Accessing and correlating context distributed across various IT and business systems is a major challenge for AI agents to function at their full potential.
- Creating a collaborative user experience where humans and agents can effectively support each other is essential for successful workflow integration.
- AI governance, control mechanisms for balancing agent autonomy with human oversight, and defining agent lifecycles are important considerations for practical implementation.
Notable quotes
*We are now here where we're looking at automating entire workflows with the use of AI SL agents and not just one task at a time.*
*The first thing that we're missing is connectors.*
*Reliability becomes a big factor and a stumbling block for us.*
Unofficial community note. Prefer the recording for nuance.