This week Seniorlink's EVP/Chief Product Officer George "GK" Kassabgi was featured in an article on Chatbots Magazine. GK discussed the chatbot within Seniorlink's care collaboration platform, Vela, and why it serves as a good example of focusing on a key functionality for a bot rather than trying to "do everything."
Via Chatbots Magazine:
Assisting medical professionals is touchy, critical, regulation-heavy work. Most chatbots don’t remember much about their customers from visit to visit. By contrast, Seniorlink’s Vela bot, which acts as a go-between to offload clinical professionals (i.e. doctors and nurses) by responding to requests and reports from home caregivers of elderly or disabled people. To do that, Vela must remember a lot about the patients themselves, to decide what advice to dispense, and when to flag the human clinician to take over — or at least make an expert decision whether to send a specific response to a specific caregiver.
Seniorlink Chief Product Officer George Kassabgi, who codes AI as well as running a startup, says most bots don’t remember much because it’s technically demanding. “To deliver a contextual conversational to the user,” he says, “a bot framework needs to incorporate longitudinal [that is, stored and reviewed across the history of conversations] data within its intents. You cannot treat that lightly, because otherwise what you’re left with is a stateless response that doesn’t and cannot care about what’s happened prior.” Not only does keeping state on patients require a profile, it requires the history of changes and new developments over time — just like your medical records.