Why Healthcare Needs AI Concierges Not Chatbots

Healthcare providers face an impossible equation: rising patient expectations, regulatory complexity, and the absolute requirement to maintain safety. You're being pitched AI solutions daily, but most come from e-commerce where "good enough" answers work.

In healthcare, there's no "good enough" when patient safety is involved.

My co-founder Jamie and I started Lorikeet because we saw this gap. Jamie came from Google Brain – he knows exactly what LLMs can and can't do safely. I came from Stripe, where precision in regulated environments wasn't optional. We built Lorikeet specifically for companies that can't afford to get it wrong.

Today, we work with leading healthcare companies like Eucalyptus (operating brands like Kin, Pilot, and Juniper across four countries), easykind in Australia, and US-based providers like nutrition health specialists Berry Street and virtual-first provider Thirty Madison. We've learned that healthcare doesn't need chatbots that deflect. It needs AI concierges that actually solve problems.

The chatbot trap

Most AI vendors push "30% automation in month 1." Sounds great until you realize this means the bot attempts 100% of tickets and fails 70% of the time. That's more than two bad AI interactions for every good one.

Healthcare support is uniquely complex. Patients don't think in neat categories e.g. they'll reply to a message about shipping confirmations with questions about chest pain. They'll ask about GLP-1 dosage in the middle of a billing inquiry. In those scenarios a chatbot that doesn't know what it doesn't know becomes dangerous.

Some vendors will claim this is your problem – that you need pristine help centers or perfect SOPs before implementing AI support. But in fact, you don't need to reorganize your entire operation, you just need AI that is built for healthcare's reality.

The concierge approach in action

Here’s how two of our healthcare subscribers have done it.

Eucalyptus operates five telehealth brands across Australia, UK, Germany, and Japan, serving 500K+ patients with everything from metabolic health to GLP-1 management. Each brand has its own voice, regulatory requirements, and clinical protocols. They needed AI that could:

• Navigate different regulatory boundaries by country

• Maintain distinct brand voices (from clinical to conversational)

• Handle complex medication queries without crossing into medical advice

• Actually resolve issues, not just acknowledge them

The results speak for themselves: 90% AI resolution on non-medical topics, human-level satisfaction scores, and 50% efficiency gains in year two. Their AI congratulates patients on health milestones while never venturing into clinical territory.

easykind launched in Australia in 2023 and immediately hit scaling challenges. Two support coordinators handling rapid growth, 24-hour response times, strict Australian compliance requirements that meant never mentioning product names. They implemented "Buddie," their AI concierge, which now:

• Reduces response times by 92% (24 hours to 2)

• Provides 24/7 support in their warm, "mate-ish" voice

• Immediately escalates when patients mention side effects

• Avoided hiring four additional staff while maintaining five-star service

Both companies expanded globally with our support, proving that compliant, high-quality AI support scales across borders.

How AI takes action in healthcare

The key difference between chatbots and concierges? Action-taking capability. Lorikeet doesn't just understand queries, it solves them through multiple approaches:

Workflows and guardrails: We combine natural language understanding with structured workflows. The AI follows your SOPs precisely when needed but can also handle edge cases conversationally. When easykind's Buddie detects a side effect mention, it doesn't guess, it follows exact escalation protocols: stop medication, call emergency services, escalate to human.

Team of agents: Healthcare support is multiplayer. A patient's delayed medication might require a telephone call to the clinic for consultation notes, confirming with the pharmacy, and updating the patient. Our Team of Agents feature spawns specialized agents to handle these third-party interactions via phone, SMS, email, or Slack. The main concierge orchestrates while specialized agents execute.

Multi-brand, multi-country: Eucalyptus's portfolio includes Juniper (weight management with GLP-1s), Kin (fertility), and Pilot (men's health). Each brand serves different demographics with distinct voices and regulatory requirements. Our AI maintains these distinctions flawlessly – a Juniper patient gets compassionate weight management support while a Pilot customer receives straightforward men's health guidance.

Building trust through control

Healthcare companies rightfully demand control. Unlike black-box solutions from companies focused on retail, Lorikeet provides complete configurability:

• Set exact boundaries: Define precisely what the AI can and cannot discuss

• Maintain compliance: easykind's Buddie never mentions product names (Australian requirement)

• Preserve brand voice: From clinical precision to warm conversation

• Track everything: Full auditability of every decision and action

This isn't about being overly cautious. It's about maximizing what AI can safely handle. When AI respects boundaries, you can confidently automate appointment scheduling, prescription refill logistics, insurance navigation, and hundreds of other non-clinical workflows.

The path forward

You've built trust with patients over years. AI should strengthen that bond, not risk it. The technology exists today to deliver true concierge-level support that:

• Knows when to act and when to escalate

• Handles multi-party coordination automatically

• Maintains your brand voice and compliance requirements

• Scales globally without compromising quality

Companies like Arbor (energy) and Linktree (creator economy) prove this approach works beyond healthcare. But healthcare is where it matters most. Your patients deserve support that solves problems 24/7, not chatbots that deflect them.

The choice isn't whether to adopt AI, it's which philosophy to embrace. Choose concierges built for healthcare's complexity, not chatbots borrowed from e-commerce.