How a Solo OB-GYN Scaled to 4,000 Questions a Month

An OB-GYN answered members one by one over Instagram DMs. An AI trained only on her 1,900+ documents now answers 92% instantly, and members made it the app's #1 feature.
4,000+
Answered every month, 24/7
Not disclosed
Implementation Time
Not disclosed
Project Cost

the challenge

Dr. Noa Sterling, a board-certified OB-GYN, built Sterling Parents in 2019 on a simple premise: pregnant people deserve better information delivered with more care. Her willingness to answer questions directly built a loyal audience, but the model had a ceiling. The core product was access to an expert's time, so growth meant either raising prices or burning out. Pregnancy generates a constant stream of urgent, personal, often frightening questions that arrive at 2 AM, not at the next appointment, and the internet's default answers skew toward the most clickbait-friendly, scary result. Before the product existed, members got answers as one-to-one Instagram DMs, sometimes waiting hours for a reply. Sterling Parents needed a way to make Dr. Sterling's expertise available to many more people, 24/7, without diluting the quality and care that defined the brand, and something that knew exactly when to say "talk to your doctor."

what they built

Sterling Parents partnered with Dewey to build Obi, a virtual pregnancy guide embedded in its website and mobile app (app built with AppIt). The name nods to the OB-GYN relationship while keeping the distinction crisp: Obi is a companion, not a clinician. Under the hood, Obi draws only from a knowledge base of 1,900+ documents published by Sterling Parents, with zero general web content or parametric knowledge, so every answer is grounded in Dr. Sterling's own expertise. 92% of questions receive an immediate, grounded answer; the system is deliberately designed not to answer everything, since in a medical domain a 100% answer rate would be a red flag. Safety guardrails reflect Dr. Sterling's philosophy: Obi never tells a pregnant person to "just relax," never surfaces unvetted speculation, warmly redirects lab-result questions to a provider, and routes medically risky questions to the care team with clear, specific language. Unanswered questions are captured as content-gap insights and aggregated so the team can prioritize new content. Because 40%+ of questions are asked by more than one member, every gap filled compounds, raising the answer rate sustainably over time.

The build started from a hard constraint: Obi could draw only on Sterling Parents' own material, never the open web or a model's parametric memory. The team assembled a knowledge base of more than 1,900 documents Dr. Sterling had already published and wired the assistant to retrieve and answer strictly from that corpus. Just as much work went into teaching Obi when not to answer. The team encoded Dr. Sterling's care philosophy into explicit guardrails: never tell a pregnant person to 'just relax,' never surface unvetted speculation, acknowledge lab-result questions warmly but hand them back to a provider, and route medically risky questions to the care team in clear, specific language. Unanswered questions became a feedback loop, captured as content-gap insights and aggregated so the team could prioritize new material against real member demand. Obi launched first on the website, then moved into a mobile app built with AppIt, where monthly sessions jumped fivefold and adoption compounded as filled gaps raised the answer rate.

best fit for

Best fit for content-rich expert businesses (solo practitioners, creators, associations, and health or professional-advice brands) whose growth is capped by one expert's time and who need to scale trusted, on-brand answers without crossing into regulated advice.

Ai ROLE
AI retrieves and composes answers strictly from Sterling Parents' own document library, and enforces safety guardrails that escalate lab-result or medically risky questions to a human provider instead of answering.

impact

4,000+ questions/month

Members' pregnancy questions answered instantly, 24/7, up from one-to-one Instagram DMs with hours-long waits.

92% answered instantly

Share of questions Obi answers immediately from Dr. Sterling's own content; it escalates the rest rather than guessing.

5x monthly sessions

Sessions jumped fivefold after the Sept 2025 app launch; Obi is the #1 feature, 4.9-star rating, with 1 in 4 members returning.

Alex Tryon

Founder
Dewey
Founder of Dewey, which builds white-labeled, closed-loop AI answer engines for content-rich organizations, trained only on their verified content and held to their compliance and brand standards.
Get an intro
Talk to this team
industry
Healthcare & Life Sciences
business organization
Customer Service
Product & Engineering
AI TYpe
Knowledge Management & Search (RAG)
Conversational AI (Chatbot / Agent)
value type
Customer Experience
Revenue Growth
Time Savings

more from this expert

No items found.

frequently asked questions

How did Sterling Parents answer 4,000 pregnancy questions a month with an AI assistant?

It built Obi, a retrieval-based pregnancy guide that answers only from more than 1,900 of Dr. Sterling's own published documents. Members get grounded answers instantly, around the clock, and Obi now handles over 4,000 questions a month at a 92% immediate-answer rate.

What AI approach powers Obi, Sterling Parents' pregnancy assistant?

Obi combines knowledge-management retrieval (RAG) with a conversational interface. It answers strictly from Sterling Parents' own library of 1,900+ documents, with no general web content or parametric model knowledge, so every response traces back to Dr. Sterling's expertise.

What results did Sterling Parents get from its AI pregnancy assistant?

Obi answers 4,000+ member questions a month at a 92% instant-answer rate, monthly sessions jumped fivefold after the mobile app launched, one in four members return as their pregnancy progresses, and the app holds a 4.9-star rating with Obi named the #1 feature.

How long did it take to roll out Obi?

Obi launched first on the Sterling Parents website, handling a modest volume through early 2025, then scaled sharply when the mobile app launched in September 2025 and monthly sessions rose fivefold.

Who is this kind of AI knowledge assistant best for?

It fits content-rich expert businesses (solo practitioners, creators, associations, and health or professional-advice brands) whose growth is limited by one expert's time and that need to scale trusted, on-brand answers without giving regulated advice.

Have a similar challenge?

Ask whether this would work for you, or describe what you're trying to solve.
TELL US WHAT YOU'RE EXPLORING