Agent Platform is Google’s managed home for AI agents — somewhere to deploy the agent you built, hand off the hosting, and get session storage, long-term memory, and scaling without standing any of it up yourself. These guides walk the whole path with one running example: a little stock-trading agent I built to exercise more of the platform than a Hello World would.
We’ll start by building that agent with ADK and deploying it to the managed runtime, then give it a memory so it remembers past runs, and finally package it as a plain container you can run anywhere — your laptop, Cloud Run, GKE, or back on Agent Platform runtime in bring-your-own-container mode. I ran each step, hit the rough edges, and show you how I got through them. Let’s dig in.
Everything in Agent Platform starts with an agent, so that’s where we’ll start too. In this part we’ll build one with ADK — Google’s open-source framework for defining agent logic, wiring up tools, and orchestrating it all in Python — and then deploy it to the Agent Platform … Continue reading Agents
Now that the trading agent is deployed, let’s give it a memory. On Agent Platform you get session persistence for free — ask it “what trades did you execute?” a day later from the same session and the whole conversation is still there, even though the agent was suspended the … Continue reading Memories
So far we’ve let other tools decide how the agent gets hosted — adk run locally, or the little deploy script that packages everything up for Agent Platform runtime. Both are handy, but sometimes you just want a plain container: one image you can run on your laptop today, and on … Continue reading Containerizing