Claude Atlassian Intelligence Claude Code Prototyping Agentic Workflows PRD · Epics · User Stories

AI in everyday product work: from backlog to a working prototype.

I use AI in two places. In the Product Owner's daily work: PRDs, epic decomposition and user stories with Claude and Atlassian Intelligence. And for prototypes: an idea often becomes something you can click through and try within a few hours, before a team invests weeks. The Product Owner stays accountable.

— Two areas

Where I use AI.

AI in the Product Owner's daily work

PRDs, epics, user stories and acceptance criteria as a draft that I review and own. It takes routine work out of the sprint and leaves more time for conversations with users and stakeholders.

AI for prototypes and real products

An idea and a PRD become a prototype that users and stakeholders can actually touch. If the assumption holds, it becomes a product with tests, security checks and acceptance. The decisions stay with me; Claude Code provides proposals, code and tests.

— From my own hands

Two examples from my own hands.

I designed and accepted both as Product Owner, built with Claude Code.

OptiSolar: live view, energy flow in real time · In production use

Ops Control Center: overview, items needing action with rule ID · Concept prototype on sample data

Ops Control Center: change, risk calculated live · 22 rules, 424 automated tests

— For whom

Is this right for you?

Product teams that spend a large part of every sprint on requirements documentation and refinement.

Heads of Product who want more output without growing the team.

Companies that want to use AI in their operations and are unsure where to start.

— What I work on

What it covers

Finding the processes where AI helps most
Prompts and workflows for PRDs, epics and user stories
From idea to prototype: clickable or working, so assumptions are tested before development
From prototype to product: tests, security and acceptance, so the experiment becomes something dependable
Connecting them to the tools the team already uses, for example Jira and Confluence
Hands-on work with the team on real tasks
A clear Definition of Done for AI-assisted work, with a person reviewing every result
Documentation, so the team can carry on without me

— Results

What I have achieved

Requirements

−30%

Less requirements overhead per sprint through AI-accelerated PRD and user story workflows. A figure from my own practice.

Prototyping

Hours

From PRD to a prototype you can try out: usually within a few hours.

FAQ

Frequently asked questions

Do we need our own data or a trained model?

No. I work with existing models such as Claude through standard tools. No model training is needed.

Do we need a data team or AI engineers?

No. This is workflow design and prompt writing, not ML engineering. Your existing product team is enough.

What if the AI produces wrong outputs?

A person reviews every result. AI delivers drafts, the Product Owner decides.

How do we measure success?

We agree on a few simple measures at the start, for example the time spent per requirements artefact, and check them again after a few weeks.

Does an AI prototype replace development?

No. A prototype answers whether something is worth building. Running it needs tests, security and someone to maintain it. With OptiSolar I planned for that from the start.

Tell me where it hurts.

A short message is enough. I'll tell you straight whether AI helps there.

Get in touch ↗