Course Objective & Scope
Not every business problem requires a fully autonomous agent; solving a predictable task with an open-ended agent loop introduces unnecessary cost, latency and unpredictability. This course maps the spectrum from deterministic LLM workflows to autonomous agents and compares core architectural patterns.
Participants implement prompt chaining, classification-based routing, parallelisation via sectioning and voting, orchestrator-workers, evaluator-optimizer (reflection) loops and autonomous tool-using agents. Teams evaluate real internal use cases and produce an Architecture Decision Record (ADR) and reference blueprint for each.
Who Should Attend?
- Solution and software architects designing enterprise AI systems
- Senior software and AI engineers building agentic applications
- Technical product managers and tech leads mapping use cases to architectures
- Prerequisite: familiarity with LLM APIs, prompting and basic tool calling
Key Takeaways
- Draw the boundary between structured agentic workflows and autonomous agents per problem
- Implement the five core workflow patterns: chaining, routing, parallelisation, orchestrator-workers and evaluator-optimizer
- Quantify trade-offs across predictability, latency, token cost and flexibility
- Define clean integration boundaries between deterministic application code and agentic components
- Produce an architectural decision matrix and reference blueprint for enterprise use cases
Curriculum
The Autonomy Spectrum: Workflow or Agent?
- Defining deterministic flows, agentic workflows and autonomous agents
- The complexity ladder: when a simple chain is enough
Structured Patterns: Chaining, Routing and Parallelisation
- Sequential prompt chains with programmatic validation gates
- Routing by intent and difficulty to specialised prompts or smaller models
- Parallel execution via task sectioning and ensemble voting
Dynamic Patterns: Orchestrator-Workers and Evaluator-Optimizer
- Runtime subtask decomposition with the orchestrator-workers pattern
- Iterative refinement loops pairing a generator with a rubric-driven evaluator
Autonomous Agents and Hybrid Architectures
- Open-ended agent loops driven by environmental feedback
- Embedding bounded autonomous nodes inside a deterministic workflow backbone
- Placing cost budgets, latency caps and safety guardrails
Architecture Design Workshop
- Mapping three real organisational scenarios to the right patterns
- Authoring an Architecture Decision Record (ADR) and component blueprint
Other Courses in This Category
-
Hands-on developer training on combining ReAct reasoning loops, Plan-and-Execute task planning and explicit state/checkpoint management into a single unified hybrid agent architecture.
- Live Course
- 5 modules
-
Technical course on delegating work across specialist agents using control handoffs versus the agent-as-tool pattern, covering triage routing, context filtering and conversation ownership.
- Live Course
- 5 modules
-
Course on designing reusable skill packages, progressive context disclosure, bundled scripts/references and dynamic capability loading so agents scale without overflowing their context window.
- Live Course
- 5 modules
-
Course on cross-framework and cross-vendor agent collaboration using the A2A (Agent2Agent) protocol: Agent Cards, task lifecycle, streaming updates, enterprise authentication and MCP–A2A complementarity.
- Live Course
- 5 modules
-
Training on pausing agents for human approval on high-risk tool calls and low-confidence decisions, editing tool arguments, routing asynchronous escalations via Slack/Teams/web UIs and maintaining audit trails.
- Live Course
- 5 modules
-
Hands-on training for developers on building ReAct agents from scratch: the Thought–Action–Observation loop, native tool calling, stop conditions, iteration limits and failure analysis.
- Live Course
- 5 modules
Get in touch
