AI that shows up like your best person,
on their best day.
Lyris designs and builds Lyris Agents — powered by MyHuman AI — for the conversations that matter most: patient intake, program onboarding, and the quiet moments people need someone steady to talk to.
Caller"I've got a pre-op appointment but I don't know what to bring."
Agent"No problem — I'll walk you through exactly what's needed for your procedure, and flag anything our staff should confirm with you directly."
Built for the places where people need to be met with care
Generic AI wasn't built for conversations like these
Most AI tools are trained on everything and grounded in nothing. When the conversation involves a patient's health, a family's application for support, or someone's hardest week, that gap shows.
One-size-fits-none scripts
Off-the-shelf bots don't know the difference between an orthopedic intake and a dental one — and it shows in every awkward question.
No sense of when to step back
Sensitive conversations need a clear line to a human. Most AI tools don't know where that line is, let alone how to hand off gracefully.
Disconnected from real expertise
A model with no grounding in your field, your clinicians, or your program's actual methodology will always sound like it's guessing.
No room to grow with you
Point solutions solve one workflow and stall. We build platforms that expand as your programs, specialties, and affiliates do.
Three places where we go deep
We don't try to be everything to everyone. Lyris focuses where AI can do the most good with the most care — and we go deep enough in each to get it right.
Patient intake that speaks your specialty's language
We build intake and scheduling voice agents trained on the specific vocabulary, forms, and triage logic of your practice — so patients feel understood from the very first call, and your front desk gets clean, complete information every time. It's centaur design at work: your clinicians' real judgment, encoded into every conversation.
🦴 Orthopedic
Pre-visit symptom mapping, injury history, and imaging prep so the visit starts with answers, not paperwork.
🦷 Dental
New patient scheduling, insurance pre-checks, and procedure prep questions handled before they ever call.
👁️ Ophthalmology
Vision history intake, symptom triage, and referral routing for urgent versus routine cases.
🩹 Pre/Post-Op Care
Pre-procedure checklists and post-op recovery check-ins that catch red flags and escalate them fast.
Caller"My incision looks a little red today."
Agent"Thanks for flagging that — I'm noting it now and a member of your care team will follow up before end of day. If it's warm, swollen, or you have a fever, call the office line right away."
Onboarding that feels like your best volunteer, not a form
We help mission-driven organizations screen, onboard, and match the people they serve — reducing wait times without losing the warmth and judgment that makes these programs work. Every agent is trained as a centaur alongside your coordinators, never around them.
🤝 Big Brothers Big Sisters
Mentor and mentee intake, match-readiness screening, and safe, consistent onboarding across chapters.
Program Partner B
Details on our next non-profit partnership are being finalized.
Coming soonProgram Partner C
Details on our next non-profit partnership are being finalized.
Coming soon Caller"I want to volunteer but I'm not sure what's required."
Agent"That's completely okay — I'll walk you through only what's needed, and a program coordinator will review your match before anything is finalized."
AI Life Support: a life coach guided by you, supported by AI
Self-guided personal support for the moments between sessions — grounded in more than 50,000 hours of real clinical practice, not generic advice pulled from the internet. You set the direction. AI Life Support helps you stay honest, stay grounded, and keep moving. This is centaur design in its purest form: one person's clinical intuition, encoded and available whenever you need it.
🧭 Grounded, not generic
Built on a real clinical framework, not a generic AI assistant persona.
🗣️ Guided by you
You lead every conversation. It reflects, asks, and grounds — it doesn't lecture.
🕰️ Always within reach
Available whenever you need to think out loud — 2pm or 2am.
🛟 Knows its limits
Built-in escalation for anything beyond self-guided support, every time.
Caller"I keep avoiding a conversation I know I need to have."
Agent"Let's slow down on that. What do you think is underneath the avoiding — is it the conversation itself, or what it might mean once it's said?"
Built on the Human-Algorithm Centaur
Researchers at Harvard Kennedy School describe the most capable AI systems today not as pure algorithms, but as "centaurs" — models that fuse machine scale with real human judgment through continuous, symbiotic learning. That's the model we build to.
Clinicians · Coordinators · Tom
Speed · Consistency · Availability
Agent
Symbiotic learning
Every agent we ship keeps learning from the real experts behind it — clinicians, program coordinators, Tom — instead of freezing at launch.
Human intuition, encoded
Tone, judgment calls, and escalation thresholds all originate from real human expertise, not generic training data scraped from the internet.
Neither alone is enough
A human alone can't be everywhere at once. An algorithm alone can't be trusted with judgment. A centaur does both, together.
Framework based on Harvard Kennedy School's research on the Human-Algorithm Centaur, applied to voice agent design.
A clear process from first call to live agent
Every engagement moves through the same disciplined process — because trust is built by knowing exactly what happens next.
Discovery
- Discovery call We learn your specialty, your programs, and where conversations break down today.
- Needs analysis We map where AI genuinely helps — and where it shouldn't be involved at all.
- Scoped proposal A clear plan, timeline, and cost before any build work begins.
→ needs analysis
→ scoped proposal
Design
- Conversation mapping We script the flows, tone, and escalation rules specific to your field.
- Safety design Sensitive-topic detection and human handoff rules are designed in from day one.
- Stakeholder review Clinicians, staff, or program leads sign off before anything is built.
→ safety design
→ stakeholder sign-off
Build & Test
- Agile build We build in stages, testing against real scenarios as we go.
- Edge-case testing We simulate the hard conversations, not just the easy ones.
- Quality assurance Security, accuracy, and tone are reviewed before anything goes live.
→ edge-case testing
→ QA review
Launch
- Phased rollout A controlled launch that limits disruption to daily operations.
- Staff training Your team learns how to review, override, and support the agent.
- Change management We stay hands-on through the first weeks live.
→ staff training
→ hands-on support
Learn & Improve
- Performance monitoring We track outcomes against the metrics that matter to you.
- Iterative improvement Regular updates as your programs and specialties evolve.
- Roadmapping A clear view of what's next as you scale.
→ iterate
→ roadmap what's next
One platform, built to handle the hard parts
Every engagement — medical, nonprofit, or personal — is built on the same centaur foundation: human expertise and algorithmic scale, learning together.
Centaur-trained agents
Grounded in your specialty's language, your program's rules, or a real clinical methodology — continuously refined by real human feedback, never a generic persona.
Human-in-the-loop safety
Sensitive-topic detection and clear escalation paths to a real person, every time it matters.
Low-latency conversation
Fast, natural responses so people never feel like they're talking to a hold queue.
Insights & reporting
Clean summaries and outcome tracking so your team always knows what's working.
Natural voice, every line
The same trained agent whether it's answering the office line, placing an outbound call, or picking up on a mobile number — never a stiff phone tree.
Compliance-aware by design
Built with HIPAA-aware and privacy-first workflows from the ground up — never bolted on later.
Designed for the conversations that require care
Lyris is built around the principle that AI should support your team and the people you serve — without ever replacing human judgment where it matters most.
Human-in-the-loop escalation on every sensitive conversation
Sensitive-topic and acute-risk detection built into every agent
Clear AI disclosure — people always know they're talking to an agent
Consent-based intake, every time, no exceptions
Staff and clinician review before final decisions are made
Full audit trails on every interaction
Data minimization — we only collect what's genuinely needed
AI never makes final eligibility, diagnostic, or clinical decisions
SOC 2 Type II
Independently audited controls for security, availability, and confidentiality.
In progressISO 42001
AI management system standard for responsible AI development and governance.
RoadmapISO 27001
Global standard for information security management and data protection.
RoadmapHIPAA-Aware Workflows
Privacy-first architecture built for protected health information from day one.
ActiveFormal certification status shown above reflects our current roadmap and will be updated as audits complete.
Common questions
Each is a distinct, deeply-scoped offering: medical and hospital intake for specialty practices, non-profit program intake for mission-driven organizations, and AI Life Support — our self-guided personal support product currently in testing. We build each on shared safety infrastructure, but the training and tone are specific to the field.
No. It's a self-guided support tool for the space between sessions, grounded in real clinical experience — not a replacement for licensed therapy, medical care, or crisis services. It's built to recognize when a conversation needs a real person, and to hand off clearly when it does.
Every medical engagement is built with HIPAA-aware, privacy-first workflows, role-based access, and data minimization from the start — we only collect what a workflow genuinely needs.
Yes — most engagements start narrow and scoped, then expand to additional specialties, affiliates, or programs once the first is live and proven.
It depends on scope, but most engagements move from discovery call to a live, tested agent in a matter of weeks, not months — with your team trained and supported through launch.
Let's build the AI that shows up for the people you serve.
Whether it's a patient, a program participant, or someone talking through their week — we'll help you build the agent that meets them well.