TippyTippy Pregnancy Companion built with Google Gemini

Source: Dev.to

This is a submission for the Built with Google Gemini: Writing Challenge

What I Built with Google Gemini

To simplify the pregnancy journey, I created TippyTippy Pregnancy Companion using Gemini and Google AI Studio, a Next.js‑based prototype app that centralizes:

• Authentication & data storage (local storage)
• Appointment & medicine scheduling
• Pregnancy timeline calculator based on LMP
• Weekly milestone details with tips and expectations
• Vitals logging (blood pressure and weight tracking)
• AI‑powered companion for insights and questions

Planned Features

• Google Health Companion integration to sync vitals from smartwatches
• Image reader for scanning prescriptions and auto‑filling details
• Firebase integration for secure data storage and cloud sync
• HIPAA compliance and other regulatory considerations
• Refined AI prompts & edge‑case handling

Demo

What I Learned

Personal Insights

• Witnessing a life grow from the size of a pea to a watermelon.
• Understanding mood swings and how everyday actions (e.g., lifting heavy objects, illness) can affect the pregnancy journey.
• The importance of empathy, positivity, and calmness during stressful moments.
• Practical caregiving skills such as holding a baby and changing diapers.

Technical Takeaways

• Prompt Engineering – Small detail changes dramatically improve Gemini’s output accuracy.
• UI/UX – Expecting parents need information at a glance; an intuitive experience is crucial, even if the prototype is still rough.
• Security – Health data is highly sensitive. I learned about encryption at rest, encryption in transit, end‑to‑end encryption, and HIPAA compliance. These considerations have delayed a public release while I continue to strengthen security.

Google Gemini Feedback

Google Gemini serves as an excellent pair programmer when given clear context, but there are areas for improvement:

• Dates – Gemini sometimes mixes different methods for calculating due dates.
• Hallucination – Occasionally it claims a code change was made when it wasn’t.
• Reinventing the wheel – The model may avoid popular npm packages and attempt to recreate functionality from scratch, both in Google AI Studio and Google Anti‑Gravity.

I mitigated these issues by explicitly reminding Gemini of the correct date methodology and using Chain‑of‑Thought prompting to reduce hallucinations.