Your developers are shipping more code with AI. But is your team actually delivering better?
A software factory supervises AI coding agents automatically, so that people can focus on higher-level work. By enforcing a consistent process for development and testing, it can significantly reduce cycle time, rework and token spend.
Vendors will try to sell you one. But for a factory to be effective, it has to encode your team’s and your organisation’s standards, processes and policies at a level of detail that has never been needed before. Growing your own factory is what makes your agents work the way your organisation works, and build products the way they should be built for you, not for someone else.
This Agentic Software Factory course guides you through a five-step process to grow your own software factory, based on research with more than 300 early adopters of AI coding tools and on several real implementations of the approach. Demonstrations run on Claude Code, but the approach applies to any agentic coding tool.
With Gojko Adzic, you’ll work through the practices and the measurements that tell you whether your team is delivering better, not just more.
The Trainer 
Gojko Adzic is the author of Impact Mapping, Specification by Example, Fifty Quick Ideas To Improve Your User Stories and a few more books.
He is the founder of Narakeet and Votito.
Gojko has helped organisations around the world improve their software delivery, from some of the largest financial institutions to small innovative startups. This workshop builds on his research with more than 300 teams adopting AI coding tools.
Learning Objectives
By the end of the workshop, you will be able to:
- Diagnose your current mode of agentic development. Recognise what makes a software factory different, position your team on the five-step adoption model, and understand what’s required, technically and organisationally, to move up from where you are.
- Plan your next step. Define concrete actions to move to the next stage of the model, including the changes it requires in CI, testing and specifications.
- Make it pay off for the team. Set up team practices and the metrics that show you’re delivering better, not just more, so that individual productivity gains become team and organizational performance.
Target Audience
This Agentic AI course is for senior developers and architects who have the authority and the responsibility to shape how their team works: internal processes, technical standards and tooling.
You should be comfortable with modern delivery practices such as CI, automated testing and working with specifications. Hands-on experience with agentic development helps, but it isn’t required.
Topics
With the help of Gojko Adzic, the workshop will cover:
- What is a software factory, and how using one looks like compared to other modes of agentic development
- Why is owning a software factory better than buying one or using open-source tools
- The adoption model: five steps to grow a factory, and how to spot where you are on the map (based on research)
- How to move from each step to the next, with concrete advice
- What this means for CI, testing, and specifications
- How to use a factory effectively as a team
- What to measure/track to ensure you’re delivering better, not just more.
- How to turn agentic work from individual productivity gain into team or organizational performance improvement.
How it works
The workshop is a mix of lectures, discussions, and demonstrations, with the group participating in a shared development session. You can, if you wish, bring your own laptop and agentic setup and follow the demonstrations to work independently, but this is not necessary.
Requirements
The ideas and methods presented are generic enough to be applied to any agentic tool or model. To keep things simple and consistent in the workshop, we will use Claude Code as an example. Although we will not cover other tools directly, you will learn enough about the underlying principles and practices to be able to apply this approach easily to any other tools after the workshop.
FAQ
- Do I need any prior knowledge to attend?
Experience with agentic development helps, but it isn’t required. You should be comfortable with modern delivery practices such as CI, automated testing and specifications. - Do I need to bring my laptop with me?
It’s optional. If you’d like to follow the demonstrations hands-on, bring your laptop with your own agentic setup. - What do I need to install?
Nothing is required. If you plan to follow the demonstrations on your own laptop, make sure your usual agentic development tools are installed and working before the workshop, with an active account or subscription.
- Should I come with my team?
It’s not required, and the workshop works well for individual participants. But since it’s about turning individual AI productivity into team performance, coming with one or two colleagues makes a real difference: you’ll share the same model and vocabulary, assess together where your team stands on the adoption map, and leave with a plan you’ve already agreed on rather than one you still have to sell to others.
