What Is Google ADK (Agent Development Kit)? A 2026 Guide
Google ADK is an open-source framework for building, evaluating, and deploying AI agents. See its components, ADK 2.0 workflows, runtime, and Atlan fit.
Kovid Rathee is the Head of Solution Architecture — Data & AI at Nexifi in Melbourne, Australia. Over 14 years he has moved from teaching at an engineering college through full-stack development, data engineering, and DevOps and cloud engineering to agentic data engineering. He writes on data engineering, context engineering, harness engineering, and the semantic layer, with bylines at The New Stack, Redpanda, QuestDB, InfluxDB, and Atlan.
Kovid Rathee is the Head of Solution Architecture — Data & AI at Nexifi in Melbourne, Australia. He advises on and builds data platforms and agentic data engineering solutions with technologies spanning all major cloud and data platforms, including AWS, Azure, Google Cloud, Databricks, Snowflake, and ClickHouse. His first job was teaching at an engineering college, from which he moved to full-stack development, data analysis, data engineering, DevOps and cloud engineering — and now he spends most of his time in the agentic engineering space.
He has worked for early-stage startups, mid-sized businesses, and large enterprises alike, building data platforms to solve business problems at scale across software services, consumer technology, finance, legal, and market research. Along the way he has worked with and led teams of data engineers and analysts, giving him the depth and breadth required to implement large-scale solutions.
Kovid has been writing technical blogs since he was studying computer engineering. He has been published with The New Stack, Atlan, Redpanda, Equinix, QuestDB, InfluxDB, and DataStax, among others, covering cloud and data engineering, relational databases, database optimization, metadata and context engineering, and harness engineering. He believes writing is the best way to refine and organise knowledge and make it useful — for oneself and for others. His passion for writing continues beyond technology: he also writes about classical music and literature.
Google ADK is an open-source framework for building, evaluating, and deploying AI agents. See its components, ADK 2.0 workflows, runtime, and Atlan fit.
Amazon Bedrock Knowledge Bases retrieve content for RAG, while a data catalog adds meaning through metadata. See where each stops and how agents use both.
Learn how the Databricks Omnigent agent runtime works: server, runner, and UI, where hosts run, what Databricks manages, and the context it cannot supply.
Omnigent controls how AI agents run, while a neutral context layer controls what they know. See how Databricks Omnigent vs. a neutral context layer compare.
Learn what the Databricks Omnigent meta-harness is, how its uniform API, policies, and server work, where it stops, and how a context layer fills the…
Learn how to connect an AI coding agent to a data warehouse MCP server. Compare vendor-managed and self-hosted options, then set auth and permission scope.
Compare Microsoft Agent Framework with Strands, Google ADK, OpenAI Agents SDK, Claude Agent SDK, LangGraph, and CrewAI on languages, protocols, and hosting.
Kovid is an independent practitioner — the Head of Solution Architecture — Data & AI at Nexifi — who contributes to Atlan under a freelance content agreement. Articles published under his byline reflect his own analysis and expertise; Atlan’s editorial team works with him on structure and clarity but does not author content attributed to him. He reviews and approves every piece published under his byline before publication. Atlan does not pay for placement of any external coverage.