Hindsight Remembers Everything. My Job Was Making Patients See It.
Our chatbot could remember three chemo cycles of a patient's life. On screen, it looked exactly like every chatbot that remembers nothing. That was the problem I had to solve. I built the interface for OnKo, a cancer-care assistant whose best feature, long-term memory, is completely invisible unless the UI shows it. This is how I designed two Streamlit dashboards around Hindsight memory, and what…
Building an invisible memory agent for a cancer-care chatbot presented a unique design challenge. The key was to make long-term memory visible to both patients and doctors, ensuring trust in the system. Three interface rules guided this design: showing memory creation moments, gating clinical data for human approval, and keeping the memory in view during conversations.
The doctor's interface presented a gate mechanism, requiring user approval before any care plan entered patient memory. Changes were editable in a table with an "Approve & Add to Care Memory" button. Approved items were saved to SQLite and Hindsight's memory layers, reflecting the memory creation act.
On the patient side, a single daily log form contained various entries like checklist items, mood cards, symptom chips, and details. Instead of storing the entire form as one entry, the system split submissions by type. Checklists appeared as separate lines, while other entries like moods and symptoms became individual memories. This approach ensured patients could see the distinct memories created from their input.
The chatbot interface displayed memory entries as if they were part of the conversation, maintaining continuity in the dialogue. A summary button generated a handover for the next doctor visit, reflecting over the patient's memory bank.
In essence, the interface transformed invisible memory into an intentional, visible process. By making memory creation, approval, and visibility central, the design fostered trust and understanding between patients and healthcare providers.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
