Key Takeaways
1. Start with the problem, not the technology.
- The primary goal of this phase is to answer the "why": why does this activity need automation? Focus on identifying the real pain points, manual steps, and time-consuming tasks in the current state.
- Real high-impact business and user needs should drive automation, not the capability of the AI technology.
2. Choose your starting point based on scope.
- Process-First: If you have an already-identified process (for example, from the Discovery Workshop), use the "As-Is Process" method. Facilitate the team to map out steps by role/lane, and use red sticky notes to describe specific, detailed challenges (for example, "Information is in multiple systems")
- Role-First: If you are exploring how to support a specific job role, use the "Mini Persona" method to anchor the problem in a human experience.
3. Focus on the objectives the persona should reach.
- When using the Mini Persona method, guide participants to define the user's key objectives (the goals) rather than their daily tasks. For example, use "Ensure quick and accurate resolution" instead of "Answer FAQs."
- The challenges identified (the pain points) must then be directly linked back to preventing the persona from achieving those objectives, providing a clear focus area for the agent.
4. Narrow the focus area through voting.
- In both the process-first and role-first approaches, use a structured voting mechanism to select the top 3-5 challenges or 1-2 objectives that the automated solution will focus on.
- This step will help to narrow the scope early in the workshop, preventing the design effort from becoming too broad or diluted, and ensuring the final agent addresses the highest-value problems.
5. Use the hybrid approach for deepest insights.
- The most Human-Centered approach is the hybrid method: start with the As-Is Process (the bird's-eye view) and follow up with a Mini Persona for the key role performing the most manual work (the zoom-in).
This combination allows the team to validate the process flow while simultaneously uncovering human-specific challenges and pain points that a single process map might miss.