The people, decisions, and handoffs around a tool determine how useful that tool becomes. Building digital care taught me to design those conditions deliberately.
At Churchome, I was hired to lead a newly created role and designed and built Pastor Chat’s digital care infrastructure from scratch using Crisp chat technology. The service connected people with human care and supported more than 85,000 practical and spiritual conversations.
This was technology-enabled care delivered by people. Its relevance to my AI work today is the operating discipline: understand the need, equip the team, define boundaries, and review the experience.
The work behind the interface
I built conversation guides, training resources, quality practices, multilingual access, and escalation pathways, equipping more than 90 staff members and 30 volunteers. A person reaching out saw a chat window. The organization needed a coordinated service behind it.
Apply that discipline to an AI use case
- Define the task: Name the work to improve and the person who benefits.
- Map the process: Identify inputs, decisions, review points, and handoffs.
- Set boundaries: Agree on appropriate data, permitted uses, and where a person must make the decision.
- Prepare the team: Provide examples, training, and a clear path when the output needs correction.
- Review the result: Compare time, quality, rework, and user experience with the starting point.
Keep the claim proportional to the evidence
In Love + Leisure, the travel agency I co-founded, I build AI-enabled itinerary and client workflows. Those applications provide a practical place to examine how AI fits the business and where human review remains necessary.
For a new engagement, I start with a focused workflow and agreed measures. That gives the team a useful basis for deciding what to expand, change, or stop.
Where could a better workflow give your team time back? We can define a focused AI use case and build the process, guidance, and review around it.
Discuss Working Together →Explore AI Implementation →