Learn to Optimise Your AI Stack - AI Realist Workshop
Only 20 places available!
If you use AI (LLMs) in your business or even in your personal projects, this workshop is a must-attend.
It is designed for people without a technical background. You do not need to understand how GPUs work or even what a token is.
What you do need to understand is how to avoid overpaying for AI without sacrificing the quality of your products or services.
To put it simply:
If you replace Opus in Claude Code with a local model, you may save money, but you could also create significant technical debt.
If you simply route every request to OpenAI, you will likely overpay by thousands of euros.
The goal is not to use the cheapest model.
The goal is to use the right model for the right task.
For example, I have been running my Iran Monitor for months at a cost of just €6 per month.
It ran 24/7, processes news continuously, and updates every minute by combining local models with cloud APIs.
€6 per month.
Had I built the same system entirely on OpenAI or Anthropic, it would have cost thousands of euros.
This workshop will teach you how to make those kinds of architectural decisions.
What you will learn
What the token economy is and why it matters.
Alternatives to the major US providers (OpenAI, Anthropic, etc.).
The trade-offs between local models and Chinese APIs, including quality, speed, privacy, and price.
What hardware investment makes sense for running local models and when it pays off.
Alternative deployment options for Chinese models, including EU-compliant cloud providers and sovereign cloud deployments.
Practical strategies for reducing AI costs without reducing quality.
Workshop Structure
The workshop lasts three hours and is split into two parts:
Theory
We cover the concepts, technologies, pricing models, deployment options, and decision-making framework.
Practice
Participants are divided into four groups of five and solve real business problems. This is also the reason why the participation is limited to 20 people!
Challenge 1: Burning Budget on AI-Assisted Coding
Your company has 100 developers using Claude Code with Anthropic subscriptions. They are heavy users, and AI costs are becoming unsustainable.
Your task is to propose a strategy that significantly reduces costs while maintaining developer productivity.
Challenge 2: New AI Product Creates Huge Cloud Bills
Your company has launched an AI-powered product that relies on OpenAI APIs. It is generating cloud bills worth thousands of euros every month.
Your goal is to reduce AI costs by 90% without significantly affecting product quality.
Challenge 3: Pricey AI Assistant
Your employees use Claude Cowork and love it because it makes them more productive. Unfortunately, even the Max plan is no longer enough, and they have already consumed half of the annual AI budget.
Your task is to optimise costs without disappointing employees or reducing productivity.
Challenge 4: Multi-Agent Workflow Burned Half a Year’s AI Budget in Five Days
Your CEO wanted agentic AI.
Now the agents are stuck in self-improvement loops and endless reasoning.
The proof of concept has already consumed half of the annual AI budget and is about to be shut down.
Your task is to redesign the architecture so the costs come under control without sacrificing quality.
Why this workshop matters now
The economics of AI are changing rapidly.
Major providers are moving toward consumption-based pricing. GitHub Copilot, Claude Code, Codex, and many other AI tools are becoming increasingly expensive.
The earlier you understand routing, local models, alternative providers, and cost optimisation, the more money your company will save.
AI is no longer just about choosing the best model.
It is about designing the most cost-effective AI stack.
📅 30 July 2026
🕔 17:00-20:00 CEST
💻 Live online
🛠️ Hands-on workshop
Register here:
https://airealist.org/workshop-the-wrong-ai-stack
Paid AI Realist subscribers receive a 20% discount using the promo code below: