Agentic Applications

2 days · Interactive workshop

Build safe agentic applications while keeping full control of your data. You decide which parts run deterministically, which tasks agents take on, and where humans need to approve their actions. We also explore the possibilities of using local models.

Who it is for

For full-stack developers and architects who want to integrate agents into their application's frontend and backend.

What you leave with

Build an agentic application with controlled workflows, tools, and backend integration.
Connect agents to a UI where people can observe, guide, and approve their actions.
Evaluate agent behaviour, test workflow boundaries, and handle failures.

Syllabus

Backend

Define the boundaries within which agents can act.

Overview of agentic frameworks, including Mastra and Microsoft Agent Framework
Deep dive into Google ADK, which we use throughout the workshop
Building workflows and defining their steps
Implementing tools and connecting them to backend APIs with explicit permissions
Handling failures, retries, and approval points without losing track of the process
Using classification models, such as Jev, for quick and cheap decisions

Quality and Security

Testing workflows and agents with evals
Tracing model calls, tool use, and workflow execution with Langfuse
Applying guardrails to agent inputs, outputs, and actions
Protection against prompt injection
Detecting and masking personal data (PII) before it reaches models or traces

Frontend

Integrating agents into your frontend for a better user experience, using open standards and protocols.

Connecting your frontend to agents with CopilotKit
Using AG-UI as an open standard for communication between agents and your frontend
Dynamically constructing UI elements with A2UI
Building human-in-the-loop controls to guide agents and approve their actions
Giving agents access to application context and sharing state
Using frontend tool calling to interact with your application
Embedding MCP Apps into your application

Starting knowledge: Experience building applications and working with APIs, plus basic familiarity with LLM calls and context. Attendance at AI Engineering is not required.

Dates and pricing

Tell us about your team, current experience, and what you want to achieve. We will discuss the scope, format, duration, and pricing with you.

You can book this workshop independently. For goals spanning several topics, explore our custom programmes.

Enquire about dates and pricing Compare the workshops →