Building Claude-Powered AI Agents (1 day)
This one-day hands-on workshop is designed for enterprise AI practitioners, IT professionals, and citizen developers who want to move beyond conversational AI to build Claude-powered agents. Participants will learn when to use Claude agents versus simple chatbots or workflow automations, and where these agents fit into enterprise systems. Through a structured build progression, delegates will design and implement a grounded agent using plugins and enterprise connectors in the morning, then extend it in the afternoon with automation workflows, multi-step reasoning, and autonomous behaviours. Governance, compliance, responsible AI principles, and cost awareness are embedded throughout. By the end of the day, participants will have built, tested, and demonstrated a practical Claude agent aligned with a real organisational use case.
Objectives
- Evaluate when to use Claude-powered agents versus chat tools or workflow automation
- Explain what Claude agents are and how they extend conversational AI
- Distinguish between grounded, autonomous, and multi-agent orchestration approaches
- Design and build a grounded Claude agent using plugins and enterprise connectors
- Configure actions, multi-step reasoning, and tool calls via the Anthropic API and MCP
- Apply prompt engineering techniques to improve reliability and control
- Design autonomous agent workflows with appropriate human oversight and guardrails
- Apply governance, compliance, cost control, and responsible AI principles to agent deployment
- Deploy, share, and manage agents securely within enterprise workflows
Prerequisites
- Familiarity with enterprise productivity applications such as Teams, Gmail, Slack, or SharePoint
- Experience using AI chat assistants in day-to-day work
- Access to a Claude or Anthropic workspace with API access and plugin activation
- No prior software development experience is required, though familiarity with basic workflows and data concepts is helpful
Contents
Strategic Foundations: Claude Agents in the Enterprise
- Differentiate Claude agents from chat assistants and workflow automation tools
- Evaluate when an agent is the right solution versus a simpler workflow or app
- Describe grounded, autonomous, and multi-agent orchestration concepts
- Understand how Claude agents integrate with enterprise plugins and APIs
- Review real-world enterprise use cases across HR, IT, customer service, and operations
Designing a Grounded Claude Agent
- Define a clear business use case and success criteria for an agent
- Access and configure enterprise plugins and connectors
- Create a grounded agent with structured system instructions and persona definition
- Apply prompt engineering techniques for reliability, transparency, and user control
- Test and iteratively refine agent behaviour using realistic scenarios
Connecting Agents to Data and Enterprise Systems
- Connect agents to enterprise data sources using plugins, APIs, and secure connectors
- Integrate third-party systems such as Salesforce, ServiceNow, Slack, or Gmail via plugins
- Evaluate data quality, access permissions, and security considerations
- Test knowledge-grounded responses and multi-step reasoning accuracy
Adding Actions, Automation, and Multi-Step Reasoning
- Distinguish between knowledge-grounded responses and task-executing actions
- Configure workflows and triggers using Anthropic API and Model Context Protocol
- Implement multi-step agent plans, tool calls, and autonomous reasoning loops
- Apply human-in-the-loop oversight and escalation paths
- Test actions end-to-end across enterprise workflows
Designing Autonomous and Multi-Agent Workflows
- Identify suitable use cases for autonomous or multi-agent orchestration
- Design agent interactions and reasoning loops for collaborative problem solving
- Configure scheduling, event triggers, and data-driven workflows
- Assess risk, token usage, and cost implications of autonomous workflows
- Monitor agent performance, outcomes, and compliance metrics
Deploying, Managing, and Governing Agents
- Deploy agents within enterprise systems and plugin-enabled environments
- Manage agent lifecycle, versioning, and rollback
- Apply governance, security, and compliance controls for enterprise deployments
- Audit agent activity and monitor performance using logs and dashboards
- Define organisational policies for responsible AI usage and agent creation
Capstone: Building and Presenting an Enterprise Claude Agent
- Define a realistic organisational use case
- Design the agent architecture, including knowledge, tools, and workflows
- Build and test the agent using plugins and multi-step workflows
- Present the agent, governance approach, and deployment plan
- Plan next steps for adoption and scaling across the enterprise
