Workshop: Agent Systems
Mentor - Danylo Topchii, AI engineer, Technical founder
We invite you to a practical workshop on creating agent systems: OpenClaw, Hermes, Paperclip.
Automate your own work and business processes offline!
You'll build your own agent-led company and learn how to identify and cover use cases.
- You will understand:
- how these systems work
- how to configure security
- how to manage a budget.
This is not a theoretical lecture about the possibilities of the LLM, but an practical workshop: you start from scratch assemble a working agent system on your own laptop or server, connect it to real tools via MCP and see how an individual agent transforms on a team of agents that interact with each other and perform tasks without you constant participation.
- A map of what processes you can automate with agents
- Understanding the Patterns Behind Agent Frameworks
- Working setup for own routines and work
What you will get at the output
Format and duration
3 sessions of 2.5 hours each
Language of the event
Ukrainian
Language of the presentation
English
Who will be interested?
- Developers of all levels
- To PMs and all who manage projects
- Anyone with minimal technical knowledge who needs to build agent systems.
Program
29.09, 18:30. Fundamental concepts, setup, security
- Agent systems architecture. Let's understand how agent frameworks act as an autonomous architecture for LLM. Understanding OpenClaw as a basic orchestrator.
- Setup and security. Setup on your computer or server. Granting permissions and security restrictions.
- Configuring your agents. We analyze the main parts of the architecture (user.md, memory.md, soul.md and agents.md) and create our own agents with their personalities, skills and memory.
1.10, 18:30. Agent Engineering, Protocols, and Processes
- Connecting agents to the outside world. Creating skills that teach agents to perform specific tasks. Connecting your agents to hundreds of MCP servers to use third-party systems.
- Agent-to-agent interaction (A2A protocol). Moving from a single agent to building a multiagent system where agents interact with each other..
- Agent memory. We'll figure out what types of memory exist, build our own memory library, and give an agent access to it.
4.10, 12:00. Scaling the System to a Real Business
- Practical system. Applying all previous knowledge to build a multi-agent system that solves tasks — starting from a morning briefing, monitoring your social media, creating content, finding clients and outreach, all the way to writing code.
- Scaling up to a full-fledged organization with Paperclip. If Hermes and OpenClaw are your workers, then Paperclip is the company itself. Setting goals, monitoring execution and project management, and managing the budget.
- Expanding memory for use within the organization. Frameworks, access restrictions for information that each user has access to.
Данило Топчий, Technical founder
— 9+ років досвіду роботи в стартапах
— Пройшов два найбільших VC акселератора – Entrepreneur First та Antler
— Автор подкасту "Потестим в проді"
—Спікер українських та міжнародних AI конференцій
— Консультант стартапів по Generative AI