An introduction to the traditional workflow where developers manually translate design specifications from tools like Figma into code.
AI is presented not as true intelligence but as a powerful tool for automating repetitive development processes and workflows.
A breakdown of key AI terms, defining LLMs as the brain, agents as the body with tools, and MCPs as the API connecting them.
An agent can generate Playwright tests from natural language prompts by following a predefined set of rules and instructions for consistency.
Setting up a local MCP server allows an AI agent to read a Figma design URL and generate corresponding component code in your IDE.
A workflow file with step-by-step instructions enables the AI to generate multiple components, styles, and Storybook stories in a single, automated run.
Figma Connect bridges the gap between design and code by syncing component properties, states, and variables, giving the AI better input for more accurate code generation.
AI tools are powerful assistants, but developers must retain control, understand the generated code, and not blindly trust the output to avoid errors.
A conversation on whether MCPs are a lasting standard, the challenges of AI's lack of memory, and the risk of over-automating development processes.
44:41
36:41Raymond Camden
39:11Thomas Dohmke
31:39Rainer Stropek
04:58Karan Shetti
28:24Emil Sjölander
30:36Clemens Helm
43:05Kenneth Auchenberg & Christian Heilmann
Jobs that call for the skills explored in this talk.
AVEX Automotive GmbH & Co. KG
Hamburg, Germany
Avanade
Berlin, Germany
Adria Solutions ltd
Manchester, United Kingdom
Avanade
München, Germany
