Building Repeatable AI Workflows with Louis Choquel, CTO @ Pipelex

# Building Repeatable AI Workflows with Louis Choquel, CTO @ Pipelex

🗓️ 9th October, 2025
🕓 4pm Irish Time

***

Agents are explorers. Workflows need engineers.
AI agents are brilliant at solving new problems, but they often struggle when asked to perform the same task repeatedly. Review 10,000 expense reports and you will get 10,000 slightly different answers. The intelligence is there, but the consistency is not.

That is where knowledge pipelines come in.

Louis Choquel, co-founder and CTO of Pipelex, believes the next breakthrough in AI isn’t a new model, it is repeatability.

In this live Ask Me Anything, we will discuss how Pipelex’s open-source framework lets developers and agents build modular, reliable AI workflows using a simple declarative language.

Think of it as Docker for AI workflows, a deterministic structure with adaptive intelligence. When a method works, you can finally capture it, version it, and share it.

***

What we will cover

• Why most agent workflows break on repeatability
• How knowledge pipelines turn ad-hoc prompts into structured systems
• What it means to build deterministic yet adaptive AI workflows
• The architecture of understanding: how pipes compose like Lego
• Live Q&A with Louis on what’s next for the Pipelex open-source project

***

About Louis Choquel
Co-Founder & CTO at Pipelex Louis leads the team behind Pipelex, an open-source framework and declarative language for information-processing pipelines using AI. The project helps developers and AI agents build modular LLM workflows and leave brittle glue code and mega-prompts behind.
Before Pipelex, Louis spent over a decade in game design and product leadership. He was involved in creating hit games such as Treasure Madness (18M downloads) and Riddle Stones (5M downloads), and later served as Product Director at Smach. His work blends creative systems thinking with deep technical insight into how humans and software learn.