Wireframe draft — for discussion, not final visual design

LillyDirect Care Coordinator

Patient and physician experience, powered by Cognesity. LillyDirect patients get continuous support between visits through Cogni; their referring physician is notified automatically and kept updated — no opt-in required. This walks through both sides of that loop.

v0.4 · reflects the Cognesity → LillyDirect proposal (07.29.26) · click a step below, or use Prev / Next
Patient journey
Physician view — different persona, referenced for context
For our discussion — not a product screen
Step 1 of 11
9:41●●●

LillyDirect Care Coordinator

POWERED BY COGNESITY

Dr. Priya Nair referred you to LillyDirect for your treatment. Your Care Coordinator keeps you supported — and Dr. Nair updated — between visits.

Activate my Care Coordinator
How this works
Screen 1 of 11

Activate

The patient has already enrolled in LillyDirect and been referred for GLP-1 therapy. This is the low-friction next step — activating the Care Coordinator, not a separate account decision.

Backend tie-in
  • Referring physician (shown here) is the same record LillyDirect already has from the GLP-1 referral — this is what lets physician notification happen automatically with zero physician-side setup (proposal p.3–4).
  • Deep-link entry from a LillyDirect referral email should land here with the physician already pre-filled, not ask the patient to re-enter it.
9:41●●●
← Back

Let's confirm it's you

We verify identity before any health information is shared — the same check Cogni runs before every check-in.

Jordan
Alvarez
MM / DD / YYYY
Your treating physician — on file, please confirm
Dr. Priya Nair
p.nair@bostonweightcare.com
(617) 555-0199
This is what we have on file — let us know if anything's changed.
Continue
Screen 2 of 11

Identity verification

Confirms the patient matches an existing chart or employer enrollment record before any clinical content is shown or discussed.

Backend tie-in
  • Mirrors the voice channel's Identity Verification state (Architecture v3 §6) — same two core fields (name + DOB), so a patient starting a check-in later is asked exactly what they already confirmed here, not re-interrogated.
  • This record becomes the patient_id key the Memory Layer (v3 §17) is built around.
  • Physician name/email/phone are new fields as of this pass — pre-filled from the existing chart, but shown and editable rather than hidden, so a stale or wrong contact doesn't silently mean summaries and alerts go nowhere. Email is what makes delivery work at all; phone is captured for potential future use (e.g. SMS alerts) even though nothing reads it yet.
Dev notes
  • Failed match should not reveal *why* it failed (existing account vs. no record) — same non-disclosure principle as the phone channel.
Viewing: First-time Steady-state
9:41●●●

Welcome, Jordan

Let's set up your care plan
A quick conversation with Cogni — covers your current medication, weight goals, and how you're feeling. About 5–10 minutes.
Start now
Medication
Not set up yet
Weight trend
Not set up yet
Message care team
Recent activity
Nothing yet — this fills in after your first check-in.
Home
Trends
Alerts
Learn
Care team
Screen 4 of 11

Home — two states, same screen

Toggle above switches between them directly, rather than describing the second one in prose. A brand-new patient has no medication or weight data yet — showing "On track" or a future-dated check-in would be false, so the first-time state prompts starting that first conversation now instead.

Backend tie-in
  • "Start now" / "Talk to Cogni now" both lead into Screen 5, same mechanism either way — but the first session is where the baseline actually gets established. Medication and Weight Loss are mandatory every check-in (v3 §9.1–9.2), and their first-call phrasing variants (built for exactly this moment) ask "What medication are you currently taking?" cold, rather than "Last time you were on X — still the case?"
  • No separate intake form exists or is planned — Cogni's own pathway questions are the single source of this data, in the exact wording the pathway tables specify. A typed form asking a similar-but-not-identical question would just create a second, uncontrolled copy of the same fields.
  • Symptoms are covered without adding a third mandatory pathway — Medication and Weight Loss stay the only two, deliberately, so calls don't bloat into a checklist. Instead, the existing TRIAGE opener (v3 §7) gets a first-call-specific version: rather than a generic "How can I help today?", the first session asks something more pointed — "How have you been feeling since starting your medication — any side effects?" — so the trigger-cue system (v3 §10) has something concrete to scan and route from on the one call where missing something matters most.
  • Timing (5–10 min) is a rough estimate, not a tested number: MED (7 questions) + WL (5 questions) + the mandatory closing sequence puts a quiet first call around 5–6 minutes; a symptom that triggers a pathway (e.g. Digestion, 6–7 more questions) pushes the long end toward 10–13. Treat this range as a placeholder to revisit once a real conversation gets timed.
  • Which state renders is driven by whether a first check-in has happened yet — not a setting the patient controls; the toggle here is a wireframe device only, not part of the product itself.
9:41●●●
← Back to Home

Checking in with Cogni

Scheduled check-in · can also be started anytime from Home

Cogni is listening
Speak naturally — Cogni waits for you to finish before responding.
End session
Screen 5 of 11

Voice session with Cogni

Corrected from earlier drafts, which modeled this as an incoming phone call — that's not right. Telephony (clinic calling a patient, or a patient calling a clinic) is Provider Portal territory. This app's voice interaction happens entirely in-app: a scheduled check-in, or started on demand from Home ("Talk to Cogni now").

Backend tie-in
  • Same Turn Cycle and pathway engine as the rest of this architecture (v3 §2, §8) — the interaction model doesn't change, only the transport does (in-app audio session vs. a phone call).
  • Scheduled entry: a push notification/reminder opens directly into this screen at the scheduled time. On-demand entry: tapping "Talk to Cogni now" on Home does the same.
  • First session specifically: TRIAGE's opener is more pointed than usual ("How have you been feeling since starting your medication — any side effects?") rather than a generic "How can I help today?" — see Screen 4's annotation for why.
9:41●●●
← Back to Home

Here's what we covered

Thursday, Aug 14 · 2:04–2:13 PM

You're taking your medication as prescribed.
Your weight has continued trending down since last month.
You mentioned some mild nausea — we'll keep an eye on that.
✓ Shared with your care team
Share this with my care team now
Ask a question about this session
Screen 6 of 11

Session summary

Plain-language recap in the patient's own app — not the clinical transcript, not a tier label, just what was actually discussed.

Backend tie-in
  • Generated from the session's answered_questions and DDM state (v3 §4), rewritten in plain language.
  • The mild-nausea line is a good example of an Escalation-Relevant field (Escalation Logic §3) that's Follow-up-weighted, not alarming — patient sees a calm reflection of it, not a clinical severity score.
  • "Shared with your care team" (green, already sent) and "Share this with my care team now" (button, patient-initiated) are two different triggers into the same delivery mechanism — one automatic per the tiering logic, one on demand. Neither depends on the provider having set up anything beyond receiving email.
9:41●●●

Care alerts

We'll only post here when your care team needs to follow up — not for routine check-ins.

Care team notified
Aug 12 — Your care team was notified about symptoms you shared and may reach out.
Follow-up completed
Aug 3 — We recommended you seek immediate care. Your care team followed up the same day.
Home
Trends
Alerts
Learn
Care team
Screen 8 of 11

Care alerts

Patient-friendly reflection of Urgent/Emergency events — deliberately does not expose internal tier names or clinical scoring.

Backend tie-in
  • Maps to the four-tier model in Escalation Logic §1/§5 — the amber card reflects an Urgent tier, the red card reflects a past Emergency escalation, written as a calm, past-tense, reassuring log entry rather than re-surfacing crisis language.
9:41●●●

For you

Suggested based on what you've shared in your check-ins.

Managing mild nausea on GLP-1 medications
4 min read
Staying on track with your dose schedule
3 min read
Understanding your weight trend
3 min read
Home
Trends
Alerts
Learn
Care team
Screen 9 of 11

Learn

Education surface — content selection can be personalized by which pathways a patient's check-ins have actually triggered, not generic content for everyone.

Backend tie-in
  • Candidate signal: completed/triggered pathways this check-in and last (v3 §10 trigger cues) — the nausea article above follows directly from a Digestion pathway trigger.
9:41●●●

Your care team

Dr. Priya Nair
Your referring physician
Last summary shared: Aug 1
Aug 2New care instructions from Dr. Nair — view
Send a message
Dr. Nair receives your updates automatically — nothing extra for either of you to set up.
Home
Trends
Alerts
Learn
Care team
Screen 10 of 11

Care team

Shows who's treating the patient and closes the loop the proposal describes: the physician is notified automatically, and can send care instructions back — this is where the patient sees that response land.

Backend tie-in
  • "New care instructions from Dr. Nair" is the patient-side landing point for the physician's "Respond to Patient" action shown in Screen 12 (the physician's Clinical Summary view).
9:41●●●
← Back

Settings

Profile
Jordan Alvarez · DOB on file · (617) 555-0148
Program
LillyDirect — referred by Dr. Priya Nair
SMS reminders
App notifications
Privacy & data
Log out
Screen 11 of 11

Settings

Standard account management, plus a visible confirmation of which program and (where applicable) which physician drove the enrollment — reads differently depending on the Screen 1 path taken.

Inbox — Dr. Priya Nair
🔴 Urgent — Jordan Alvarez
Reported a symptom meeting urgent-review criteria — for your awareness.
LillyDirect Care Coordinator
Clinical Summary: Jordan Alvarez — open below
LillyDirect Care Coordinator — Clinical Summary
Patient: Jordan Alvarez  |  DOB: on file  |  Prescriber: Dr. Priya Nair
Jordan Alvarez has connected with you through Cogni's encrypted email updates — this is how you'll receive their care summaries and any urgent alerts going forward.
Current medication / dose
Tirzepatide 5 mg — Week 7 of titration
Adherence score
92% (11 of 12 doses)
Weight trend
↓ 9.4 lbs since initiation
Side effects
Mild nausea, improving
Engagement
8 of 8 check-ins completed
Tier
Follow-up
Suggested discussion topic
Consider reviewing readiness for escalation to 7.5 mg at next visit.
No urgent alerts this period — Jordan will be escalated immediately if severe symptoms are reported
View Jordan Alvarez's Progress Portal →
Respond to Patient →
This summary supplements — not replaces — clinical judgment. No diagnosis or treatment change is recommended by the LillyDirect Care Coordinator.
Screen 12 — physician view

Clinical summary & urgent alert

Built directly from the proposal's own example (p.5–6) — reused the same patient (Jordan) and physician (Dr. Nair) from the patient-side screens so the two views read as one continuous story. "AI Risk Score" (the proposal's own field) is now replaced with our actual internal tier — resolved, not just relabeled.

Backend tie-in
  • Tier badge now uses our real vocabulary: Emergency, Urgent, Follow-up, Routine (Escalation Logic §1, §5) — color-coded consistently across this deck: red = Emergency, amber = Urgent, blue = Follow-up, green = Routine. This example is Follow-up: nothing urgent, but the nausea + dose-escalation timing is worth a look.
  • The urgent inbox item (left) is the Urgent tier's near-real-time alert (Escalation Logic §5) — independent of the physician ever opening it, matching the proposal's central safety claim: "patient safety escalation is independent of physician review." Jordan already got directed to appropriate care before this email was even sent.
  • The teal banner above replaces what was a separate "Provider Onboarding" screen in an earlier draft. There's no distinct onboarding step here — the physician's actual first contact is LillyDirect's own existing referral/notification channel (outside this product's scope), and this banner is simply the first thing they see once Cogni's own encrypted-email updates start.
Not a product screen

Discussion Topics for Our Design Workshop

Per the design workshop already proposed as a next step — five things we'd like to align on together before finalizing. Each links back to the screen it came from.

Screen 3 — Check-in preferences
Consent & disclosure language
Exact wording for check-in consent will need a pass from your legal/compliance team before this ships.
Screen 8 — Care alerts
Escalation tier wording
Patient-safe phrasing for each alert tier is a first pass from us — worth a review from your clinical/medical affairs team.
Screen 9 — Learn
Education content source
Should this be LillyDirect-provided content, co-branded, or sourced independently?
Screen 12 — Clinical summary & alert
Provider Portal scope
Is deeper physician access — full patient timelines via the Progress Portal — part of this phase, or a later one?
Screen 12 — Clinical summary & alert
Safety-independence disclaimer
Exact wording confirming patient safety guidance never waits on physician review needs your legal/clinical sign-off.
Draft wireframes for internal review and partner discussion — layout and copy are illustrative, not final. Patient screens (1–11), the physician view (12), and our discussion topics (13) are intentionally kept low-fidelity so the conversation stays on flow and content, not visual polish. Built against the Cognesity → LillyDirect proposal, 07.29.26.