Design the workflow before you write a prompt
A ten-step method for designing AI workflows, from splitting the work into steps to the hooks that keep agents in check.
A practitioner's guide · Edition 1.0
From Business Case to Production Agents
A complete framework for taking an AI application from idea to production, in 14 components: the business case, agent design, data, evaluation, interfaces, storage, deployment, security, monitoring, cost, fine-tuning and LLMOps. Every chapter comes with worked examples, decision tables and checklists.
Watch
Seven steps, fourteen components, and an application that keeps improving.
Inside the book
Each component is a chapter, a short video and a blog post.
Is AI the right tool, and is it worth it?
Design the workflow before you write a prompt.
Trusted data in, before any model sees it.
Know that it works, and keep knowing.
Interfaces people trust and can correct.
The right store for every kind of data.
Ship the app anywhere, the same way.
Where the model runs, and why.
Safe, compliant and in control.
See every request, error and resource.
See what the agent did, and why.
Quality per euro, by design.
When prompting is not enough.
Every change versioned, tested and reversible.
From the blog
A ten-step method for designing AI workflows, from splitting the work into steps to the hooks that keep agents in check.
Before choosing a model, define the problem, its cost and how you will know it worked. Many tasks are best solved with ordinary code.
Golden datasets, graders and evaluation gates that turn "it looked fine in the demo" into measurable quality.
Get the full book with all 14 chapters, diagrams, checklists and worked examples.