AI-Hands-on

Practical AI build patterns for SMEs.

Implementation-oriented topics for operators, product owners, and automation leads who want to see how AI-assisted software can be built, tested, and governed in real workflows.

Implementation playbooks

These guides focus on practical workflow design: inputs, automation steps, control points, and metrics that show whether AI is helping.

A2A and MCP enterprise architecture demo video thumbnail AI-Hands-on 13 · Enterprise architecture

A2A + MCP enterprise architecture, layer by layer

Combining discovery, reasoning, routing, tooling, streaming, and deployment into one production-shaped framework.

A2A Vision Agent deployed with Docker and Cloud Run demo video thumbnail AI-Hands-on 12 · Cloud deployment

Deploying an A2A Vision Agent with Docker and Cloud Run

Containerizing a multimodal agent and shipping it to a real public Cloud Run endpoint.

A2A official SDK with LangChain streaming demo video thumbnail AI-Hands-on 11 · Real-time AI

Real-time streaming with the official A2A SDK and LangChain

Replacing one blocking JSON reply with a live working / input_required / completed event stream.

A2A combined with MCP tooling demo video thumbnail AI-Hands-on 10 · Agent tooling

Combining A2A with MCP for direct system access

One Host Agent, two tool types: A2A for talking to other agents, MCP for reading files and running local tools.

A2A multi-agent LLM routing demo video thumbnail AI-Hands-on 09 · Multi-agent orchestration

Multi-agent routing with an LLM-driven Orchestrator

Three single-purpose agents plus a registry and an Orchestrator that lets an LLM decide who handles each request.

A2A agent upgraded with Google ADK and Gemini demo video thumbnail AI-Hands-on 08 · AI agent protocols

Upgrading an A2A agent with Google ADK and Gemini

Turning a hard-coded reply into a real reasoning agent with session memory and JSON-RPC 2.0.

A2A protocol discovery and task execution demo video thumbnail AI-Hands-on 07 · AI agent protocols

A2A protocol basics: agent discovery and task execution

How two AI agents that have never met still understand each other, via a shared discovery and task-execution handshake.

n8n workflow architecture for a lead qualification agent using strict JSON and routing AI-Hands-on 06 · Deep tech workflow

Building a production-ready n8n lead qualification agent

Study a real n8n + Claude workflow and turn it into a stronger production pattern for scoring, routing, replying, and logging leads.

Operational automation pipeline for cross-platform data synchronization AI-Hands-on 05 · Integration

Building an agentic middleware for messy cross-platform data sync

Use agents only around the messy parts of integration: validation, deduplication, explanation, and exception handling.

AI combines CRM pipeline data and team updates into a verified weekly sales report AI-Hands-on 04 · Sales reporting

Automating weekly sales reports from CRM and team updates with AI

Create a repeatable sales brief that combines pipeline movement, team context, source links, and risk signals.

Field voice note transformed into a reviewed structured company wiki entry AI-Hands-on 03 · Voice workflow

Turning field operator voice notes into structured company wiki entries

Capture field knowledge as reviewed, searchable documentation instead of scattered recordings.

Template workflow adapted into an onboarding process AI-Hands-on 02 · HR workflow

Building an automated onboarding workflow with AI document routing

Turn one new-hire intake into the right HR, IT, finance, manager, and compliance tasks.

AI invoice extraction workflow from email attachments to validation approval and ERP update AI-Hands-on 01 · Finance documents

How to automate invoice extraction from email to ERP with AI

Move invoices from inbox to extraction, validation, approval, audit, and ERP update with fewer manual touches.