Feedback topics
Designing Officevibe’s first AI product, shaping strategy and delivering a high-impact reporting experience.
Context
As organizations scale, leaders face an overwhelming volume of employee feedback. Managers and HR often return from vacation to find dozens of pages of unread comments, making it nearly impossible to keep up. At the same time, HR leaders aren’t always interested in individual pieces of feedback, they need to understand the bigger picture: which topics employees are talking about most, and the sentiment behind them.
At Officevibe, this presented a dual opportunity. On the business side, we needed to position ourselves as an AI-first product. For our customers, solving feedback overwhelm meant stronger engagement, more participation, and higher retention. We defined success by tracking report visits, actions taken after reviewing a topic, and changes in sentiment over time.
At Officevibe, this presented a dual opportunity. On the business side, we needed to position ourselves as an AI-first product. For our customers, solving feedback overwhelm meant stronger engagement, more participation, and higher retention. We defined success by tracking report visits, actions taken after reviewing a topic, and changes in sentiment over time.
Process
We began by speaking with customers about how they handled feedback today. Many were exporting CSV files and running their own AI analysis outside of Officevibe, losing valuable context in the process.
We explored multiple directions. One was an AI agent that would guide users to the most pressing issues. But while powerful, it felt like a black box and required familiarity with AI tools. Instead, we leaned toward a more open and approachable solution: an AI-generated report that would surface the most important themes and trends in a way managers already understood.
I led the design strategy within our product trio. I framed the problem, defined the design principles, and worked with engineering and data science to scope a lean but high-value solution.
We explored multiple directions. One was an AI agent that would guide users to the most pressing issues. But while powerful, it felt like a black box and required familiarity with AI tools. Instead, we leaned toward a more open and approachable solution: an AI-generated report that would surface the most important themes and trends in a way managers already understood.
I led the design strategy within our product trio. I framed the problem, defined the design principles, and worked with engineering and data science to scope a lean but high-value solution.
Solution
The final product was an AI-powered feedback report. It tracked the overall volume of feedback and organized it into topics. Each topic could be explored in depth: users could see whether sentiment was positive or constructive, discover related themes, and even reply directly to specific comments.
The interaction was designed around focus and intention. Users could move from a high-level overview of trends, into a single topic, and finally into specific feedback that required action. By highlighting sentiment shifts and surfacing the most actionable comments, we enabled managers and HR to go from trend → insight → action with minimal friction.
We made tradeoffs along the way. For example, AI wasn’t free to invent categories, we predefined a taxonomy of over 400 topics to keep results consistent and interpretable. This meant some feedback didn’t fit neatly into a category, but it gave users confidence in the structure of the report.
The interaction was designed around focus and intention. Users could move from a high-level overview of trends, into a single topic, and finally into specific feedback that required action. By highlighting sentiment shifts and surfacing the most actionable comments, we enabled managers and HR to go from trend → insight → action with minimal friction.
We made tradeoffs along the way. For example, AI wasn’t free to invent categories, we predefined a taxonomy of over 400 topics to keep results consistent and interpretable. This meant some feedback didn’t fit neatly into a category, but it gave users confidence in the structure of the report.
Impact & learnings
The report changed how our customers managed feedback. HR leaders, who typically deal with the highest volume, could now understand at a glance what employees were talking about and which topics needed immediate attention. Team managers, often unsure how to prioritize replies, could now focus and act fast.
This was Officevibe’s first AI project. Beyond delivering a valuable product, it helped us learn what was possible, what customers were ready for, and where we could push further. Future opportunities include more visual exploration of feedback and highlighting urgent comments within each topic.
For me personally, this project was a crash course in designing with AI. I learned how to balance the openness of machine learning with the clarity users expect, and how to position AI not as a black box, but as a transparent, approachable assistant that drives meaningful action.
This was Officevibe’s first AI project. Beyond delivering a valuable product, it helped us learn what was possible, what customers were ready for, and where we could push further. Future opportunities include more visual exploration of feedback and highlighting urgent comments within each topic.
For me personally, this project was a crash course in designing with AI. I learned how to balance the openness of machine learning with the clarity users expect, and how to position AI not as a black box, but as a transparent, approachable assistant that drives meaningful action.