2019 Specialisation: backend (python & Go), data engineering and agentic AI Hobbies: anime, games e cosplay Blog: danicat.dev LinkedIn: linkedin.com/in/petruzalek Twitter: x.com/danicat83 Google Cloud 02
e achieve objectives on behalf of its users. Basically: a model with access to tools (functions) that allow it to interact with the external world Google Cloud 07
(PS2) for £15? user prompt + system prompt + tools Tool call request: query_database("puyo puyo") query_database("puyo puyo") { "found": ["puyo puyo tsu (PS2)"]} Tool call response { "found": ["puyo puyo tsu (PS2)"]} Final response: "No, you already have it" Google Cloud Final response: "No, you already have it" 09
Abstraction level: low (basically an 1:1 mapping of the APIs) Supports: Generate Content API, Live API e Batch API Auth via Gemini API and Gemini Enterprise (Vertex AI) github.com/googleapis/go-genai Alternative: ai.google.dev/gemini-api/docs/openai Google Cloud 011
the plugin system Abstraction level: medium Primitives: flows and prompts (dotprompt) Middleware: supports tools, skills, etc. Official skills: npx skills add genkit-ai/skills Also available for JS/TS, Python and Dart genkit.dev Google Cloud 013
other labs Abstraction level: high Primitives: agents Official skills: uvx google-agents-cli setup Also available for JS/TS, Python, Java and Kotlin adk.dev Google Cloud 016
simple agents. Agent run as a container that responds to requests over HTTP Gemini Enterprise Agent Platform Unified platform to build, deploy, govern, and optimize enterprise-grade AI agents and model-based solutions. Use when leveraging enterprise-grade products like Memory Bank, Sessions, Registry, Gateway, etc. Google Cloud 018