

Linux Festival · 2026
The OpenSourceAgentStack
What changes when software starts to reason, and the open protocols that keep agents interoperable.

Mehdi Ghalenoei
AI Engineer at System Group
About me
Mehdi Ghalenoei
AI Engineer at System Group
My path
- Frontend Dev
- Software Engineer
- DevOps Team Lead
- AI Engineer
mehdigh@systemgroup.net
Today
From ERP to open standards
The system we know
Architecture before agents
From clicks to goals
What changes when software chooses its steps?
Build with discipline
Scope · design · build · evaluate · observe
Connect the ecosystem
MCP · A2A · A2UI · AGENTS.md
Part 1
An ERP system
Part 1
Software development life cycle (SDLC)
Part 1
A reasoning layer joins the ERP
Part 1
Agents are moving into production
of teams already run AI agents in production
LangChain, State of Agent Engineering, Dec 2025
of large companies now scale AI agents, in one year
McKinsey, The State of AI in 2026, Aug 2026
skipped buying software because coding agents could build it
McKinsey, The State of AI in 2026, Aug 2026
of agent projects may be cancelled by 2027
Gartner, Jun 2025
Part 2
An LLM talks. An agent gets work done.
get_stock()Part 2
Anatomy of an agent
Goal
“Do we have enough laptops?”
Think
Check stock and sales.
Act
get_stock("LAP14")
Observe
{"stock": 12, "weekly": 30}
Answer
“No. 3 days left. PO drafted.”
Part 2
Four principles of an agent
Goal-driven
Works toward a goal
Autonomy
Decides how
Reactivity
Adapts to change
Proactivity
Acts before problems
Part 2
Why agents are hard to build
Part 2
The probabilistic shift
Part 3
The Agent Development Lifecycle
Give it a goal and rules
Eight design decisions
Workflow
Fixed steps or a flexible graph?
Patterns
One agent or a team?
Context
What does the model see?
Tools
Which ERP functions, with which rights?
Framework
Which framework to build on?
Model & cost
Which model, at what price per task?
Evaluation
How do we know it works?
Observability
How do we see what it did?
Workflows
Agentic patterns
Router
Like a receptionist: sends each request to the right expert.
Reflection
A second agent checks the work before it goes out.
Agentic patterns
Human-in-the-loop
A person approves risky actions before they run.
Group chat
Agents contribute to one shared thread; the manager selects who speaks next.
Multi-agent architecture
Context engineering
Framework, model & tool selection
Which LLM?
Model
Which framework?
Framework
What can it touch?
Tools
Shift the mindset: test the what and the how
The new QA pipeline
Observability: AgentOps
DevOps · the body
Is the server OK?
AgentOps · the mind
Is it thinking well?
“Why is invoice 4711 blocked?” ├─ get_invoice(4711) ✓ ├─ get_invoice(4711) ⚠ repeated ├─ check_policy() ✓ pass └─ answer ✓ grounded
A healthy server can still hide a confused agent.
Part 3
The open-source toolbox
Part 4
Open governance for the agent ecosystem
“The internet runs on open protocols and open source. So will agentic AI.”
aaif.io
agent ↔ tools
from Anthropicagent ↔ agent
from Googlenotes for coding agents
from OpenAIPart 4 · Open standards
AGENTS.md: project instructions for coding agents
A simple, open format for guiding coding agents, used by over 60k open-source projects.
## Project Use the existing ERP service patterns. ## Checks Run `pnpm test` before proposing changes. ## Conventions Keep API contracts backward compatible. Never commit secrets or customer data.
Part 4
How the standards work together
Part 4
MCP: a USB-C port for AI
MCP SDK downloads in one month
npm, Sep 2026
Model-controlled
Tools
App-controlled
Resources
User-controlled
Prompts
Part 4
A2A: agents talk to agents
organisations support A2A
Linux Foundation, Apr 2026
Part 4