#117 Designing for AI at Miro
Honest UX Talks
Deep Dive
Why is research crucial in designing AI features?
Research is essential because the AI industry often replicates models without questioning their effectiveness. Without research, we can't challenge existing standards or determine if they're suitable for our specific problem spaces. It helps avoid building the wrong models or blindly following trends like ChatGPT.
What challenges did Ioana face when joining Miro as an AI designer?
Ioana faced multiple layers of ambiguity, including understanding the company's processes, team dynamics, and the vast amount of existing research and concepts. She also had to navigate the uncertainty of designing AI features for a collaborative tool like Miro, where the interaction models were still being invented.
How does Ioana approach designing AI features to ensure they solve real problems?
Ioana emphasizes avoiding AI for the sake of AI and instead focuses on solving specific user problems. She ensures that AI features are intentional and address real friction points, such as the cold start problem in Miro's canvas, where users feel overwhelmed by the freedom of a blank space.
What is the 'cold start problem' in Miro, and how does AI help solve it?
The cold start problem refers to users feeling overwhelmed by the vast possibilities of a blank canvas in Miro. AI helps by providing an entry point, like the 'Create with AI' feature, which guides users on where to start and reduces the initial intimidation of the canvas.
How many of Ioana's initial AI concepts made it into production at Miro?
Ioana explains that while many concepts were explored, only those that were technically feasible and provided a good user experience made it to production. Some forward-thinking concepts remain in the abstract stage due to technical limitations or the need for more research before they can be implemented.
What is the process for developing an AI feature at Miro?
The process involves balancing immediate technical feasibility with forward-thinking innovation. Designers must consider user problems, technical capabilities, and the future potential of AI. Prototyping and building real features are necessary to learn how they perform, as traditional prototyping methods are insufficient for AI.
How does user research influence the development of AI-driven features at Miro?
User research is crucial for understanding how AI features are perceived and used. It helps validate or invalidate assumptions about the value and usability of AI features. However, research can also be ambiguous, with contradictory insights from users, requiring continuous questioning and refinement.
What is the role of Sidekicks in Miro's AI features?
Sidekicks are AI entities that provide expertise in specific roles, like an agile coach or product leader. They help users overcome knowledge gaps in their teams and guide them through workflows. Sidekicks are designed to be easy to understand and use, with a clear value proposition.
What is the biggest challenge in designing AI features for Miro?
The biggest challenge is balancing immediate technical feasibility with long-term innovation. Designers must think about what can be built today while also envisioning future possibilities that may not be technically feasible yet. This requires constant abstraction and tactical thinking.
What advice does Ioana give to designers working in the AI space?
Ioana advises designers to focus on asking the right questions rather than jumping to solutions. The AI space is highly uncertain, and confidence in a solution often means you're missing something. Continuous research and iteration are essential to building meaningful AI experiences.
- Initial focus on understanding Miro's internal workings, existing research, and general AI experiences.
- Emphasis on solving actual user pain points, avoiding AI for AI's sake.
- Iterative process of design, implementation, and user feedback.
- Balancing immediate needs with long-term vision for AI integration.
- Importance of measuring the impact of AI features and reflecting on their performance.
Shownotes Transcript