Ela AI Agent

Member acquisition AI Agent for DPCs

DPC model of care is fairly new in US. DPC website & phone number are largest business channels for DPC practices. Member signing up encounter lot of concerns & questions. Since websites are not designed to clarify & convince— these visitors drop off & DPCs lose revenue. Ela is an end to end membership growth engine for DPC. Ela engages website visitors & incoming callers instantly, explain your membership services in friendly way, clears any initial doubts, builds trust, & sells memberships — even after office hours.

Client

Supercare

Roles & Responsibilities

Market Research
JTBD analysis
Interaction & Visuals
Prototyping & Conversational Testing

Users

DPC potential members
DPC Front desk

Devices

Responsive Web

The $250m problem

Patient drop-offs are digging a $250m+ hole in DPC economy, every year.

Drop-off rate

~80

Conversion rate

~20

First response

12-24

hrs

70–80% of visitors drop off within 2 minutes. Each drop-off is lost revenue. With memberships worth $1,500–$2,000 a year, even 50 missed sign-ups equals $75,000–$100,000 in lost revenue annually.

Qualitative Surveys

To uncover the reasons behind drop-off, we conducted qualitative surveys with patients who recently attempted to purchase membership

Flagged patterns like mis-trust, confusion etc

Multiple pages, FAQs & redirections

Anxiety regarding loss of Insurance benefits

The trust to action journey

Websites are most authentic foot prints of DPC business and are crucial to patients purchase decisions

Identifying pain signals

We conducted JTBD surveys to uncover top pain signals in the process

Job To Be Done	Visitor Intention (Why they came)	Importance (1–5)	Satisfaction (1–5)	Gap (Pain Signal)	Insight for Ela
Understand what DPC really means	“I want to know if DPC can replace my insurance.”	5	2	+3	Visitors don’t understand model — need conversational explainer, not paragraph.
Find if clinic is right for me	“Does this clinic fit my family, budget, or chronic care?”	5	2	+3	No personalization — static content fails to adapt to visitor type.
Feel trust before sharing info	“Can I safely share my number or medical needs?”	5	1	+4	No trust signals or response loops — need human-like assurance.
Book a call or appointment easily	“I just want to talk to someone.”	4	2	+2	Booking feels bureaucratic — Ela can automate instant confirmation.
Learn pricing clearly	“How much does it actually cost monthly?”	5	3	+2	Pricing pages are vague — Ela can decode contextually.
Ask questions without feeling judged	“I don’t want to sound dumb asking about care.”	4	1	+3	Websites can’t offer emotional safety — Ela can.
Get a callback or answer fast	“If I leave a message, will they reply soon?”	5	2	+3	70% forms go unanswered — Ela bridges delay instantly.
Feel the clinic’s personality	“Does this place feel warm, human, and modern?”	4	2	+2	Sites feel sterile — Ela adds empathy and story.
Complete forms or assessments painlessly	“I don’t have time for 15 questions.”	3	2	+1	Friction in onboarding — Ela converts forms into chat flows.
Gain confidence to sign up	“Am I making a smart choice?”	5	2	+3	Emotional conversion missing — Ela becomes the confidence bridge.

Distilling user archetypes

We distilled our findings in to 5 personas / archetypes for efficient empathy mapping

The foundational shift

I realized that reducing patient drop-offs required more than just a better interface — it needed a shift toward intent-driven engagement. I led a workshop aligning business, clinical, and technology teams, using data to show that most drop-offs stemmed from delayed responses and weak trust signals. I proposed investing in an always-on AI agent instead of adding staff or traditional forms, which led to the AI initiative being prioritized over other roadmap items. This decision ultimately reduced projected support costs per conversion by 27%.

The Ela messenger for website

Ela messenger works with website and enables user to do most important jobs for patient— Greet visitors, explain DPC via voice, book appointment, purchase memberships, or request call back to name a few.

Mini-widgets for messenger

Ela messenger comes with mini widgets that can scale Ela’s capabilities

Ela control centre for clinic users

The getting started experience that handholds user to setup essentials for Ela

Train & Personalise Ela

Ela’s performance dashboard lets you track Ela’s performance along with new visitors attended everyday

While Ela handles all the visitor conversations, clinic user can take control anytime. Ela also smartly hands-off when required saving 2-3 mins per patient everyday.

Ela onboards the patient and notifies clinic with a summary of activities it conducted for patient.

Ela books appointment request and brings to clinic staff for confirmation.

Ela brings summary and sentimental analysis. Ela self rates the conversation and shows the CX score too.

Ela's validation & testing

Ela Scores

1132

Fin Scores

965 Gap

17

We used parallel response comparison methodology with with 8 comprehensive metrics, scoring each response from real screenshots and conversations. The questions were divided and Ela was put to test against real-world patient scenarios

15+ test scenarios

Membership questions
Emergency situations
Service assistance
Cultural awareness
Non-member support

Scoring system

Max per metric (150 points)
Evaluation (Parallel comparison)
Total (1200 points)

Key focus area

Patient empathy & care
Clarity of communication
Emergency handling
Conversion optimisation
Cultural sensitivity

Clarity of communication

Ela gives to the point answers. Ready for takeaway. While having a human like response give by Fin is desirable, our focus was to make answers quickly consumable

Emergency situation

While Ela clearly knows where to step back and avoid providing medical assistance, Ela also knows the essentials that she should assist with. It provides first aid actions. However, it emphasises to call 911 immediately.

Cultural aptitude

Ela understands the cultural convenience and helps identify suitable team mate for further course boosting the convenience and comfort

Fixing user feedback

Most users perceived Ela as just another support widget typically ignored on websites

During our guerrilla testing with 6–7 participants from diverse age groups, we noticed a consistent behavior — most users perceived Ela as just another support widget typically ignored on websites. This insight made it clear that Ela’s current positioning blended into the background rather than standing out as an intelligent, proactive presence. To address this, we decided to strategically distinguish Ela across key aspects such as her visual identity, conversational behavior, and interaction placement — ensuring users instantly recognize her as more than a support tool, but as an active, intelligent care assistant.