Coding agents make building software faster and cheaper. What matters is how teams use the capacity this frees up. The Agentic Trio extends Teresa Torres' Product Trio to include delivery and invests that capacity in better discovery instead of more output. The approach was developed by Daniel Westheide and Fabian Walther and is laid out in 12 theses.
Agentic Trio
Using coding agents to build the right thing
We’d love to assist you in your digitalization efforts from start to finish. Please do not hesitate to contact us.
Get in touch!
Frequently Asked Questions
Do you have questions about the agentic trio? Here you will find answers to questions we are frequently asked.
What is the Agentic Trio?
The Agentic Trio is an approach for product teams that work with coding agents. Product management, UX and engineering share responsibility for both discovery and delivery: deciding what should be built, and building and shipping it. The goal is better outcomes for users and organisations, not more generated code. The term was coined by Daniel Westheide and Fabian Walther at INNOQ.
All 12 theses on the Agentic Trio: agentictrio.org
How is the Agentic Trio different from the Product Trio?
In Teresa Torres' Product Trio, a small unit of product management, UX and engineering jointly figures out which problems users have and which solutions are worth pursuing. Delivery has been handled by the rest of the product team. Coding agents make implementation faster and cheaper that the same three roles can now take on delivery as well. That’s why the Agentic Trio has no delivery-only roles.
More in the blog post The Agentic Trio and in Thesis 6
How do the roles of product management, UX and engineering change?
All three roles work with coding agents, and all three take part in discovery. Their skills converge, but each role keeps its own focus. Product management and UX build solutions with agents themselves and need to be able to judge what generated code can do. Engineering remains the role with technical depth and makes sure the team can keep delivering safely and productively.
What is an agent harness?
The agent harness is the framework coding agents work within. It includes guardrails such as architecture checks and linter rules, skills and subagents for recurring tasks, a CI/CD pipeline that every piece of generated code has to pass through, and observability. In the Agentic Trio, engineering builds and maintains the harness. It has to be in place before the trio gets started with agents, otherwise technical debt piles up quickly.
More in Thesis 9, Thesis 10 and the post Building to Understand
Does the Agentic Trio fit every organisation?
No, the Agentic Trio is not a universal blueprint. It is one answer to the question of how faster delivery can lead to better outcomes. It works best when the trio has real decision-making autonomy and can make use of short cycles. It doesn’t compete with Team Topologies either: Team Topologies describes how teams are structured, while the Agentic Trio describes how discovery and delivery are organised within a team.
More in the blog post The Agentic Trio