AI Assistant

Context

In the competitive healthcare industry, there is a growing demand for technology solutions that help doctors reduce administrative burden and improve patient care. Our product, an AI-powered assistant, aimed to do just that by transcribing doctor-patient conversations and automatically generating summaries. The initial business case for the product was clear: save doctors time, allow them to focus more on patient empathy, and reduce their administrative workload. However, the product faced challenges such as low adoption rates and trust issues with AI-generated summaries, which directly impacted its success in the market.

Opportunities

The business impact was significant: if the AI assistant could help doctors spend less time on documentation and more time with patients, it would not only improve patient satisfaction but also lead to increased efficiency, higher patient capacity, and ultimately a stronger market position for the company. In my role as Senior Product Designer, I was tasked with addressing these challenges by optimizing the user experience to drive both user adoption and business growth

Problem

The AI assistant was designed to transcribe and summarise doctor-patient visits. However, it faced several hurdles that hindered its success:

• Low adoption rate (35%): Doctors were hesitant to rely on AI, preferring manual note-taking instead.

• High rejection of AI summaries (42%): AI-generated notes required frequent edits, causing doctors to lose trust in the product.

• Increased visit times: AI’s inefficiency in producing accurate summaries led to an average of 5 extra minutes per visit, limiting doctors’ ability to see more patients.

• Backend instability: Constant updates and lack of a clear vision resulted in a fragmented user experience.

Goals

  • Increase adoption rate

  • Decrease summary rejection

  • Decrease time spent reviewing summaries

User story mapping to overcome internal challenges

While designing faced significant internal hurdles:

Lack of vision and backend instability: The product was often unstable, requiring constant patching from the development team. I helped the team align by using user story mapping to create a clear, data-driven vision that prioritized features with the most impact on user experience and business outcomes.

Stakeholder-driven feature requests: Stakeholders often requested features based on feedback from just one or two doctors, which didn’t reflect the broader user base. I brought data to the table to challenge these requests and focus on real user needs backed by metrics and research.

As the lead product designer, I played a pivotal role in shaping the AI assistant's user experience. My responsibilities included:

  • User Research: Conducting in-depth interviews with doctors and patients to understand their pain points and needs.

  • Information Architecture: Defining the information hierarchy and user flows to ensure a seamless and intuitive user experience.

  • UI/UX Design: Creating visually appealing and user-friendly interfaces that aligned with DocPlanner's brand guidelines.

  • Prototyping and Testing: Iteratively designing and testing prototypes to validate design decisions and optimise the user experience.

  • Collaboration: Working closely with a cross-functional team of developers, product managers, and stakeholders to bring the product to life.

Solution and design Iterations

1️⃣ Mobile Recording – Enabling Seamless Workflow:

I redesigned the experience for recording visits by enabling doctors to use their mobile devices as mics, allowing them to move freely around the room while the AI captured the conversation. The recording would automatically sync to the summary page on their desktop once the visit was completed.

Impact:

• Increased adoption rate by 80% (from 35% to 63%) as doctors began using the mobile mic feature for more flexibility and efficiency.

• Saved 3 minutes per patient on average, increasing daily patient capacity by 15%.

2️⃣ Templates Page – Improving AI Accuracy & Efficiency:

I worked on improving the Templates Page, making it customisable for different types of visits (e.g., routine check-ups, follow-ups). The goal was to structure the input so that the AI could generate more relevant and accurate summaries, reducing editing time.

Impact:

  • Summary rejection rate decreased by 57% (from 42% to 18%), as the AI-generated summaries became more accurate and tailored to each visit type.

  • 40% reduction in time spent editing summaries, allowing doctors to focus more on patient care.

3️⃣ Summary Page – Enhancing Readability & Trust:

I redesigned the Summary Page to improve readability and efficiency, ensuring doctors could quickly review, edit, and utilize AI-generated visit summaries. The new layout included a clear “Edit Summary” button for quick modifications, a “Change Layout” feature for flexible viewing options, and a “Copy to Clipboard” button for seamless sharing. Additionally, I introduced a sidebar navigation for instant access to past summaries, reducing friction in follow-up visits.

Impact:

  • Doctors spent 35% less time reviewing and modifying summaries, improving workflow efficiency.

  • AI trust increased by 22%, with more doctors relying on AI-generated summaries with minimal edits.

  • System utilization grew by 15%, as more summaries were actively referenced and integrated into patient records.

Solution

Conclusion

By focusing on user-centered design and aligning the product features with real user needs, I was able to drive significant business results:

• Increased adoption and higher user satisfaction: Doctors began to trust the AI assistant for summarizing visits and documenting patient information.

• Improved product efficiency: The redesigned Templates and Summary Pages reduced the time doctors spent on documentation, allowing them to see more patients and enhance their workflow.

• Strengthened competitive position: The product became a market leader in AI-assisted documentation, boosting customer retention and attracting new clients.

Key takeaways

• Design decisions should always be rooted in user research and data—not just anecdotal feedback.

• Aligning stakeholders and building a shared vision through tools like user story mapping ensures that the team works towards a common goal.

• Product design is not just about aesthetics but about solving real business problems—whether it’s increasing adoption, reducing friction, or enhancing customer retention..

Business results and Key takeaways

This project involved developing the AI assistant as a standalone web app, as well as integrating the feature into our SaaS platform and mobile app for existing customers. However, the SaaS and mobile app implementations aren’t covered in this case study (happy to share more details in person).

The goal of these implementations was to encourage customers on the Saas lower-tier plans to upgrade to the VIP plan, which offers additional features, including the AI assistant. While the SaaS and mobile app versions were focused on serving our existing customers, the standalone web app was designed to attract new users. It offered limited summaries per month for free, with an upgrade required to access unlimited usage. The ultimate aim was to drive users towards adopting the Doctoralia CRM SaaS, which integrates the AI assistant with patient records for seamless use.

Below is an overview of the uptake across all implementations of the AI assistant within the Doctoralia product suite:

The standalone app demonstrated steady growth in adoption, particularly during its monetisation rollout in Germany, the first test market

Customers uptake

Last day of the quarter and we not only reached but exceeded our 900 active doctors target for the quarter

STDA (standalone app) already represents ¼ of our customer base, and is the main responsible for our growth

Number of sessions

We reached 85% of our target of 11,000 sessions