Scott Eising, Chief Product Officer, Reimagine Care
We’ve spent a lot of time this past year thinking about what it actually takes for a patient to trust an AI in their cancer care journey. Not trust it conceptually, trust it enough to describe a symptom at 2am, to follow its instructions, to believe that someone on the other end is paying attention.
A new Salesforce survey of more than 3,200 patients just put data behind something we’ve built toward from the beginning: patients aren’t resistant to AI in healthcare. They’re resistant to AI that’s not integrated with other aspects of their care.
Patients are three times more likely to trust an AI agent when it’s embedded in a clinical system rather than offered as a public chatbot. When I read that, I didn’t think “interesting finding.” I thought: that’s exactly why we built Remi the way we did.
The friction problem is real. In oncology, it’s urgent
The survey documents what anyone working in healthcare already knows firsthand. Forty-six percent of patients delay care because the digital process is too confusing. Fifty-eight percent skip necessary care because scheduling is too difficult. Two in three have run out of medication waiting for a prescription refill to be approved.
In most specialties, those are frustrating statistics. In oncology, they’re dangerous ones.
Missed check-ins create downstream risk. Skipping a symptom check-in, whether by telehealth or in person, lets a problem escalate before anyone catches it. Skipping a medication compliance check-in can delay oral oncolytic refills, risking treatment gaps or a patient falling off protocol, and can delay refills of the supportive medications tied to that check-in as well.
That gap is what Reimagine Care’s Virtual Care Center (VCC) and clinical programs were built to close. Our VCC operates as a 24/7 extension of the oncology care team, is staffed by oncology-trained nurses who are there when the clinic is closed and the patient’s concern can’t wait until morning. Remi, our AI purpose-built for oncology, is what makes scaling this model possible.
What patients are actually asking for
The survey findings are worth sitting with. Sixty-seven percent of patients say they’d rather have 24/7 AI help than wait on hold during office hours. But nearly 9 in 10 say a clear “escalate to human” option is essential before they’ll trust AI support and 90% say the same for anything touching clinical guidance.
Those two findings aren’t in tension. They’re the same finding. Patients want always-on support and they want a human in the loop. They’re not choosing between AI and people, they’re telling us exactly how AI needs to be designed to earn a place in their care.
That’s why Remi and the VCC are designed to work as one system. Remi proactively monitors symptoms between appointments, triages patient concerns, supports medication adherence, and routes to a VCC clinician the moment the situation warrants it. The escalation path isn’t a fallback it’s an intentional design decision. It’s what makes a cancer patient willing to engage at 11pm instead of waiting, worrying, or ending up in the ER.
Why “purpose-built for oncology” isn’t a marketing phrase
Patients’ top concerns about AI in healthcare are accuracy first, data privacy second. Thirty-six percent cite diagnostic or treatment accuracy as their primary worry. That number makes complete sense to me as a product builder and it’s why I’m skeptical of general-purpose AI being deployed in oncology contexts.
Oncology is complex and highly personalized. The symptom patterns matter. The treatment protocols matter. The difference between a side effect that needs monitoring and one that needs intervention matters, and it’s not something a generalized AI product without the right guardrails backed by oncology experts is well-positioned to navigate.
When Remi flags a concern, it’s doing so within oncology-specific pathways. When it escalates to the VCC, it’s handing off to a nurse who understands that context too. That chain, AI designed for oncology patients, escalating to clinicians trained in oncology, and operating inside a governed framework is what closes the gap between “AI that sounds helpful” and “AI a cancer patient can actually rely on.”
What the research means for where this goes next
Forty-four percent of patients say a 24/7 agentic assistant would make them more likely to stay within a provider’s network for follow-up care. In oncology, continuity isn’t just a loyalty metric; it’s a care quality metric. Patients who stay engaged with their care team between appointments have better outcomes. That’s not a soft benefit. It’s the point.
Sixty-five percent of patients with long-term conditions say round-the-clock digital support would make managing their health significantly easier. For a cancer patient navigating treatment, that number represents something real: a person who wants help and would engage with it if the technology were worthy of their trust.
Building product worthy of that trust is the only thing that matters in this work. The Salesforce data defines the conditions: embedded in a clinical system, governed by the provider, transparent about its limits, with a human always reachable. That’s not a constraint on what AI can do in oncology. That’s the design.
The bar is high. It should be.
Scott Eising is Chief Product Officer at Reimagine Care, where he leads the development of Remi — AI purpose-built for oncology. Reimagine Care’s Virtual Care Center (VCC) delivers 24/7 symptom monitoring, patient support, and human-escalated clinical triage for cancer patients and their care teams. Learn more at reimaginecare.com.