Arcturus Labs

Cohort 1

In much of the public discourse, RAG has been almost synonymous with Semantic Search. And while Semantic Search is a reasonable choice for plenty of RAG applications, it's often overkill. In this post, I present Roaming RAG, a novel form of Agentic RAG, which lets the agent navigate the corpus in much the same way that a human would – by reading the table of contents. This means that you don't even need retrieval – no embedding models, no infrastructure, no ETL.

Roaming RAG – RAG without the Vector Database
 

Chat interfaces are trending everywhere, but they have a fundamental flaw: your users are working on something, and when that work mixes into conversation, they waste time scrolling to find their content. Artifacts solve this by separating conversation from work products, creating a more efficient workflow where users can focus on their actual work rather than hunting through chat history. This post explores the Artifacts experience with implementation guidance and a demonstration sandbox

Cut the Chit-Chat with Artifacts

You’ve built the demo. It works — sort of. But the second you try to shape it into something real, it breaks. You change the task, tweak the workflow, or ask for something custom... and you find that you're stuck. That’s not your fault — it’s what happens when you build on borrowed patterns without understanding the core ideas underneath. This course fixes that. You’ll learn how LLMs actually work — not just how to prompt them, but how to design, combine, and control them across apps, workflows, agents, and more. And you won’t just learn the theory — you’ll build with it. By the end, you’ll be able to take an idea and confidently shape it into the LLM-powered tool you actually want to create.

Forget one-size-fits-all. Build AI apps as flexible as your imagination.

Course syllabus

Courses content covering everything in the syllabus above and providing the class with a homework assignment for the week.

Thursdays

Take time build your own project based upon upon the content we cover that week in the course.

Weekly Projects

Show-and-tell for the progress with the project assignments, and Q&amp;A related to the assignment and the course content.

Tuesdays

Course schedule

Engineers who want to cut through the noise, understand how LLMs really work, and build flexible, reliable applications with confidence.

PMs who want to harness LLMs for real product wins – and stop wasting time on hype, dead ends, and guessing what’s actually possible.

Team leads driving AI initiatives who need more than demos – they need to ship production-worthy AI apps that can continuously improve.

Who is this course for

What you’ll get out of this course

Understand how completion models really work, and how reinforcement learning shaped the capabilities we now use in chat and agents. Build the intuition that lets you wield LLMs intentionally — not just hope they behave.

Master the Core Mechanics of LLMs

Learn the essential techniques behind zero-shot, few-shot, and advanced prompting strategies. You'll see how to steer models reliably, even in real-world, messy tasks.

Control LLM Behavior with Better Prompts

RAG is the full chain from understanding user intent, crafting smart queries, pulling the best sources, and generating grounded, high-quality outputs. You'll learn how to design each step so your applications are accurate, reliable, and aligned with what users actually need.

Build RAG Systems That Truly Understand and Ground

Build conversational agents and workflow-driven agents that stay on track and complete non-trivial tasks. Learn the design patterns that make agents useful, flexible, and powerful.

Design Smart AI Agents

Master the techniques for offline evaluation, online observability, and iterative improvement — so you’re not flying blind as your app scales.

Evaluate, Test, and Continuously Improve Your Applications

When prompting isn't enough, customization is key. Learn when fine-tuning is worth it – and when it's not! Land how to use lightweight LoRA techniques to teach new behaviors or specialize your models.

Customize LLMs with Fine-Tuning and LoRA

Stay ahead by understanding the next frontier: multimodal models, new interaction paradigms, and what’s coming next in AI development.

Get Ready for the Future of AI

Drowning in AI trends? Learn the timeless principles behind LLM applications — and build with clarity, not chaos.

John Berryman

Discovering the Power of Lexical Search in RAG Applications

AI Consultant @ Arcturus Labs

You'll see the places where semantic search is weak. These are the exact places where lexical is strong.

Semantic vs Lexical

We'll explain lexical search so that you can understand how it works and why it is ideal for certain domains.

Lexical Search Basics

Attendees will get a super fast tour of how to build an end-to-end Elasticsearch-based RAG application.

Learn directly from Arcturus Labs

Arcturus Labs

Courses

Building LLM Applications - From Principles to Practice

Lightning Lessons

Discovering the Power of Lexical Search in RAG Applications

Resources

Roaming RAG – RAG without the Vector Database

Cut the Chit-Chat with Artifacts