What the talk is about The AI hype is forcing teams to make most things AI centric. We were in a similar situation too. Business wanted us to build an Agentic workflow. The sessions walk us through the use cases where one or the other is more useful.
Outline
- Introduction and Context
- The growing adoption of agentic workflows
- Why agentic systems often become the default recommendation
- The business and technical context behind our architectural decision
- Workflows
- What are workflows
- High level Classic Workflows and Agentic workflows
- Evaluating the Use Case
- A sample use case
- Identifying known steps, decision points, and outcomes
- Technical Trade-offs
- Reliability and predictability
- State management and recoverability
- Observability and debugging
- Cost and latency
- Failure handling and operational complexity
- When to Use Which Approach
- Use cases suited to classic orchestration
- Use cases suited to agentic workflows
- Decision criteria and practical design heuristics
- Common pitfalls in choosing an execution model
- Some existing workflows engines.
- Hybrid Workflow Architectures
- Combining deterministic orchestration with agentic components
- Using agents for interpretation, planning, or judgment
- Keeping state, retries, approvals, and controls in the orchestrator
- Closing Takeaways
- Choosing autonomy based on the problem, not the trend
- Using orchestration for control flow and agents for judgment
