Understand AI agents
Distinguish an assistant, a fixed workflow and an agent. Explore tool calling, context, state and the decisions an agent should leave to a person.
Agentic Lab Academy · by QPillars
Build AI agents for lab instruments.
Hands-on AI-agent training on real lab hardware.
Connect a laboratory instrument to your copilot. Build your own assistant and AI agent. Learn by developing, running and troubleshooting a practical lab workflow.
In-person workshops · Zurich & Basel, Switzerland
The agent–instrument loop you will explore
Prepare a liquid-handling workflow
Goal → agent → instrument → observations → next action
First cohorts in preparation2 daysUp to 6 participantsHands-on exercisesTaught in EnglishBasic Python
What you will learn
Understand agents by building them.
Start with what an AI agent actually is. Learn how a model, instructions, knowledge, tools and feedback work together, then put those pieces into practice on a laboratory workflow.
Connect your copilot to an instrument
Expose supported instrument operations through MCP and explore using them from a compatible copilot, such as Claude or an assistant in VS Code.
Build and evaluate your own agent
Add instrument knowledge, coordinate tool calls, observe results and troubleshoot failures. Explore where simulation, digital twins and human approval improve the workflow.
Your hands-on projects
A connection. An assistant. Your own agent.
Work in three pairs, with individual development environments and scheduled turns on real lab hardware. Guided exercises and prepared starter code help you make progress throughout the two days.
A copilot–instrument connection
Make instrument tools available to a compatible copilot and inspect the state and results it receives.
An instrument assistant
Give an assistant relevant procedures and instrument knowledge so it can help with a defined laboratory task.
An agent prototype to take home
Build an agent that proposes steps, calls tools and responds to feedback. Leave with your starter project, exercises and an evaluation checklist.
Explore digital twins as one way to test proposed actions before execution. Their checks cover the modeled constraints; supervised instrument practice helps you examine what actually happens.
Two days, one complete learning journey
From agent fundamentals to your own lab agent.
Short preparation
Before we meet
- Check your environment with a guided setup exercise.
- Bring one nonconfidential lab task you would like to explore.
Day 1
Understand agents & connect your copilot
- Explore AI agents, assistants, tool calling and the agent feedback loop.
- Connect supported instrument capabilities through MCP and explore the role of SiLA 2.
- Use those tools from a compatible copilot and build an instrument assistant.
Day 2
Build, run & troubleshoot your agent
- Build your agent around a defined lab task, with tools, knowledge and feedback.
- Run guided exercises on real lab hardware and investigate failed steps.
- Explore evaluations, human approval and the role of simulation and digital twins.
After the course
Keep moving
- Use the starter project and evaluation checklist at work.
- Join a shared follow-up clinic to discuss your next steps.
Built for laboratory builders
For the people connecting AI to the lab.
- Laboratory automation engineers
- Instrument software & application engineers
- Scientists with scripting experience
Learn together. Connect across companies.
We curate small cohorts of up to six participants, aiming to bring together people from different companies and laboratory backgrounds. Build agents together, exchange ideas and strengthen the lab automation community.
Is this course right for you?
You can read and adapt basic Python scripts and are familiar with a laboratory workflow. You do not need prior experience with an AI-agent framework. Bring a laptop; setup guidance will follow before the course.
Learning with your team?
Tell us if you are exploring a private workshop for your instrument or automation team. We can discuss the right learning scope together.
Discuss a team workshop
Learn with the engineer building it
Iacob Marian
Founder & Technical Lead, QPillars
Iacob builds software for scientific instruments, from instrument control and liquid handling to AI agents and executable workflow models. The course brings that engineering experience into exercises you can inspect and adapt.
The first cohorts
Choose where you would like to join.
We are preparing the same two-day workshop for Basel and Zurich. Tell us which cohort suits you; we will follow up with dates, venue and booking details.
Planned cohort
Basel
December 2026
In person · up to 6 participants
Express interestPlanned cohort
Zurich
January 2027
In person · up to 6 participants
Express interestIndicative course fee
CHF 2,000excluding VAT · per participant · 2 daysPlanned inclusions: guided practicals, course materials, starter project and a shared follow-up clinic. The fee excludes VAT. Final fee, dates and instrument setup will be confirmed before booking.
Join the first cohorts
Build lab agents. Meet your peers.
Interested in joining us in Basel or Zurich? Leave your details and we’ll be in touch about the upcoming courses.
A few questions before you join.
Will we work with a real instrument?
Yes. You will build and test AI agents through guided exercises on real lab hardware, working in pairs with supervised instrument time.
Do I need experience with agentic frameworks or MCP?
No. Basic Python and familiarity with laboratory workflows are enough. You learn the concepts through guided exercises.
Is the course taught in German or English?
The first courses are taught in English. We also plan to offer courses in German.
Can my employer fund my place?
The course is designed for professional development. We can provide a course overview for your manager or training team to support an internal funding request.