Uploaded November 2026 | Updated November 2026, 5 days ago
Connect AI agent teams with the Agent-to-Agent (A2A) protocol. This tutorial compares two approaches for exposing, discovering, and invoking database-backed AI agent teams: a self-managed A2A server using Oracle Select AI for Python and a fully managed A2A server in Oracle Autonomous AI Database.
See how A2A clients retrieve an agent card, send natural-language tasks, and receive results from Select AI agent teams. The walkthrough uses loan and population teams to demonstrate A2A discovery and requests with curl commands, a local Streamlit client, bearer-token authentication, and OAuth client authentication.
You will learn when to choose a standalone A2A server versus a managed endpoint, how A2A and MCP address complementary parts of an agent architecture, and how to make existing natural-language-to-SQL agent teams available to compatible client applications.
Chapters:
0:00 A2A protocol and AI agent team overview
3:42 Select AI Python A2A server documentation
4:55 Review existing Select AI agent teams
8:35 Start and query standalone A2A servers
13:47 Enable and use the managed A2A server
Key links:
A2A Protocol
a2a-protocol.org
A2A GitHub Project
github.com/a2aproject
Autonomous AI Database A2A Server overview:
blogs.oracle.com/developers/introducing-oracle-autonomous-ai-database-a2a-server-for-governed-multi-agent-systems
Autonomous AI Database A2A Server documentation:
docs.oracle.com/en-us/iaas/autonomous-database-serverless/doc/expose-database-ai-agents-agent2agent-protocol.html
Select AI for Python A2A documentation:
oracle.github.io/python-select-ai/user_guide/a2a.html
Explore more Code Innovate videos:
youtube.com/playlist?list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a
For developers building governed multi-agent applications, this video shows how A2A-compatible clients discover database-backed agent teams through agent cards and invoke them through standard A2A endpoints.
#codeinnovate #A2A #AIAgents #MCP
Connect AI agent teams with the Agent-to-Agent (A2A) protocol. This tutorial compares two approaches for exposing, discovering, and invoking database-backed AI agent teams: a self-managed A2A server using Oracle Select AI for Python and a fully managed A2A server in Oracle Autonomous AI Database.
See how A2A clients retrieve an agent card, send natural-language tasks, and receive results from Select AI agent teams. The walkthrough uses loan and population teams to demonstrate A2A discovery and requests with curl commands, a local Streamlit client, bearer-token authentication, and OAuth client authentication.
You will learn when to choose a standalone A2A server versus a managed endpoint, how A2A and MCP address complementary parts of an agent architecture, and how to make existing natural-language-to-SQL agent teams available to compatible client applications.
Chapters:
0:00 A2A protocol and AI agent team overview
3:42 Select AI Python A2A server documentation
4:55 Review existing Select AI agent teams
8:35 Start and query standalone A2A servers
13:47 Enable and use the managed A2A server
Key links:
A2A Protocol
a2a-protocol.org
A2A GitHub Project
github.com/a2aproject
Autonomous AI Database A2A Server overview:
blogs.oracle.com/developers/introducing-oracle-autonomous-ai-database-a2a-server-for-governed-multi-agent-systems
Autonomous AI Database A2A Server documentation:
docs.oracle.com/en-us/iaas/autonomous-database-serverless/doc/expose-database-ai-agents-agent2agent-protocol.html
Select AI for Python A2A documentation:
oracle.github.io/python-select-ai/user_guide/a2a.html
Explore more Code Innovate videos:
youtube.com/playlist?list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a
For developers building governed multi-agent applications, this video shows how A2A-compatible clients discover database-backed agent teams through agent cards and invoke them through standard A2A endpoints.
#codeinnovate #A2A #AIAgents #MCP










