Turning product operations into an automated decision system.
Product teams depended on repeated checks, messages, follow-ups and task routing. I designed and built an operational workflow that automated the repeatable parts and left the judgement with people.
- The problem
- A recurring manual checking process, plus the messages, follow-ups and task allocation around it, consumed product team time.
- What I did
- Designed and built the workflow while working as UX/UI Design Lead.
- What was built
- An operational workflow with automated checks, automated Product Manager messages, automated follow-ups and weighted task allocation.
- The result
- Reduced a recurring manual checking process from approximately one hour to five minutes while automating messaging, follow-ups and weighted task allocation.
- My responsibility
- Workflow design, automation, process design and implementation.
- Status
- Implemented internal workflow.
The problem
Product teams depended on repeated checks, messages, follow-ups, task routing and allocation decisions. None of it was difficult, but all of it needed someone to remember, check and send, every time.
The opportunity
Separate the work that needs human judgement from the work that is simply repeatable. Automate the second group, and make sure the first group reaches the right person with the right context.
The workflow
Manual checks, messages, waiting and follow-ups are intercepted and collapse into one automated timeline.
- Incoming taskWork enters the process.
- Automated checksThe checking that was previously done by hand is validated automatically.
- Weighted task allocationAllocation uses weighted logic rather than ad hoc assignment.
- Automated Product Manager messageThe relevant person is told automatically.
- Automated follow-upsFollow-ups no longer depend on someone remembering to send them.
Simplified recreation of the implemented logic.
Result
~1 hour→~5 min
Recurring operational checking, with messaging, follow-ups and weighted task allocation automated alongside it.
Before
Manual checking. Manual messaging. Manual follow-ups.
After
Automated checks and messaging. Automated follow-ups. Weighted allocation logic. Fewer manual errors.
AI adoption and governance
Alongside product design leadership, I worked with Legal and Marketing on the foundations of practical AI governance and internal AI adoption.
Wider AI enablement work
Alongside the workflow above, I introduced AI-assisted research, documentation and prompt workflows into the design and product operations process. Competitor and benchmark analysis for page audits was accelerated using structured Claude prompts, with human judgement kept at the interpretation, recommendation and prioritisation stages.
The aim was an organisational capability rather than personal productivity: reusable prompt workflows for audits, Jira ticket drafting, component documentation and accessibility risk identification that anyone on the team could run. I also used n8n to connect some of the repeated coordination steps between tools.
A well-engineered prompt is a governance document. It encodes what good looks like, what to check and what format the output should take. That is why it belongs in a shared workflow, not in one person's chat history.
Proposed concepts
Beyond what was implemented, I outlined several AI agent concepts for design operations. These are proposals, not live systems.
- UX governance agent: flags governance gaps, missing documentation and accessibility failures before development starts.
- Design health reporting: aggregates component adoption and audit results into a regular report.
- Content governance agent: reviews editorial content against tone, SEO and regulatory sensitivity before publication.