Hands-On Agentic Coding with Google Antigravity
Overview
AI coding is moving beyond autocomplete and isolated code generation. Developers can now delegate complete engineering tasks to agents but reliable results still depend on the quality of the context, specifications, constraints, and verification provided by the developer.
In this hands-on workshop, you’ll use Google Antigravity to complete an end-to-end agentic development workflow inside an existing application.
You’ll begin by exploring an unfamiliar codebase and turning a feature request into a structured implementation specification. You’ll define acceptance criteria, constraints, and edge cases before reviewing the agent’s proposed plan. You’ll then guide the agent as it implements the feature across multiple application layers, generates tests, diagnoses failures, and verifies the final behaviour.
Along the way, you’ll learn how to identify implementation drift, decide when to delegate and when to intervene, and introduce practical safeguards around permissions, commands, credentials, and sensitive data.
This is not a workshop about asking AI to generate more code. It is about learning how to manage an agent through a disciplined engineering workflow:
Understand → Specify → Plan → Implement → Test → Verify
What you’ll build and practise
During the workshop, you’ll work through a realistic feature-development task using Google Antigravity. You’ll practise how to:
Analyse an unfamiliar repository and identify the relevant application components
Convert a feature idea into a lightweight implementation specification
Define requirements, constraints, edge cases, and acceptance criteria
Provide an agent with the project context it needs—without overwhelming it
Review an agent-generated implementation plan before any code is changed
Coordinate changes across the frontend, backend, tests, and configuration
Detect when implementation has drifted from the agreed specification
Generate tests directly from acceptance criteria
Use the agent to investigate failures and resolve defects
Verify application behaviour through automated tests and browser inspection
Introduce repository-level instructions and reusable development standards
Apply practical guardrails for safer agent-driven development
Meet Your Instructor
Daniel Vaughan
Google Developer Expert Daniel Vaughan
Daniel helps organisations turn AI investment into measurable results by building production-ready AI systems and agentic workflows that improve productivity and quality. He leads HCLTech AI Labs’ global Forward Deployed Engineering practice, delivering embedded AI engineering teams that take solutions from prototype to production across Anthropic, OpenAI, and Google technologies.
By the end of the workshop, you’ll be able to
Explain how agentic coding differs from conventional AI coding assistance
Use Google Antigravity to explore and work within an existing codebase
Structure development tasks so agents can execute them more reliably
Evaluate and improve an agent’s implementation plan
Delegate multi-step engineering work while retaining control over key decisions
Recognise and correct implementation drift before it becomes costly
Use specifications and acceptance criteria to guide testing and verification
Establish safer boundaries for agent permissions, commands, and data access
Create reusable repository instructions for consistent agent behaviour
Apply a repeatable agentic workflow to future development tasks
You’ll walk away with
Hands-on experience completing a realistic development task with Google Antigravity
A reusable workflow for specification-driven agentic coding
A practical framework for reviewing agent-generated plans and changes
Techniques for coordinating work across multiple application layers
A clearer understanding of when to delegate, intervene, or stop an agent
Testing, debugging, and browser-based verification experience
Practical guardrails for using coding agents in real development environments
Live access to the instructor for questions and implementation guidance
Who should attend?
This workshop is designed for:
Software developers exploring agentic coding workflows
Full-stack developers working across frontend and backend systems
Engineers already using AI coding tools who want to move beyond autocomplete
Developers interested in specification-driven development
Technical leads evaluating coding agents for their teams
Senior developers and architects establishing AI-assisted engineering standards
Engineering teams looking to improve the reliability of AI-generated code
Attendees should have a basic understanding of software development, source-code repositories, and application testing. Previous experience with Google Antigravity is not required.
You’ll also receive
Full workshop recording
Certificate of completion
Live Q&A with Daniel Vaughan