Overview
Zenvia is a Brazilian SaaS platform that offers multichannel customer service solutions, helping support teams centralize communication and boost productivity. In our product Servir, we identified an opportunity to integrate AI into the support experience, going beyond the chatbots already available in our portfolio.
As a member of the Product team, I led the discovery and ideation phases, with the goal of aligning AI solutions with the company’s strategic objectives. I also supported the development and post-launch monitoring to ensure continuous value delivery.
Discovery Process
We kicked off the project with an in-depth discovery process to understand our users’ needs and expectations regarding AI in customer support. The research was conducted across multiple fronts:
- Desk Research: analysis of existing internal data and reports
- Quantitative Research: surveys to measure user interest and perception of AI-powered features
- Stakeholder Interviews: conversations with internal teams and clients to gather insights and expectations
- User Journey Mapping: detailed analysis of agents’ day-to-day workflows to identify major pain points
- Benchmarking: research into how other market players were leveraging AI in their support solutions
Key findings from our research included:
- 60% of our user base already expressed interest in AI functionalities
- Agents frequently lost time getting up to speed with long tickets and searching for scattered information
- Many were already using external AI tools to assist with writing
Expectations were clear:
- For users: reduce response time and improve support quality
- For the company: increase NPS and protect MRR
Based on these insights, we defined the core AI features for our support experience:
- Agent-facing insights
- Request summarization
- Response suggestions
- Information retrieval
- Writing assistant
Ideation & Prototyping
With the functionalities defined, we moved on to the ideation phase, which included:
- Solution Design: exploration of different approaches for each feature
- Business Validation: alignment with company strategy
- Design Critique: collaborative sessions with the design, content, and design system teams
- Usability Testing: conducted with users who participated in the discovery phase
- Technical Feasibility: validation with engineering to ensure we could implement the solution
- Handoff: detailed documentation of the experience for development
Outcome & Delivery
After several iterations and rounds of user feedback, we delivered a robust tool that integrates AI into the support workflow. The results were significant:
- 38% increase in product NPS
- 30% reduction in average ticket resolution time
- 9% MRR retention, by preventing churn from at-risk customers
In the first three months:
- The AI analysis features (insights, summarization, and response suggestions) were triggered 317,000 times
- The writing assistant was used 15,000 times
These numbers reflect strong adoption by support agents and confirm the value of the solution in handling complex tickets more efficiently.
Key Learnings & Final Thoughts
This project was a valuable opportunity to apply research and user-centered design methods to solve real problems. It reinforced the importance of deeply understanding user needs and the power of cross-functional collaboration in building impactful solutions.
I believe that the ability to listen to users and work collaboratively is essential to designing products that not only meet customer needs but also drive business outcomes.