GlucoCoachAI iOS & Android, live Jay Loyola was diagnosed with type 2 diabetes and left with a stack of pamphlets. He wanted an AI coach that answers before you eat, not after. AppMakers USA designed and built GlucoCoachAI end to end, on iOS and Android, and shipped it to both stores.
Jay Loyola was diagnosed with type 2 diabetes in January 2025. His doctor gave him some medication, a few pamphlets, and a line he still repeats: change your diet, and I will see you in three months. He was, in his words, left in the dark.
Every app he tried was reactive. He would log a meal after eating it, and the app would tell him what he should have done. Jay wanted the opposite: an AI coach he could ask before he ordered, before he cooked, in the moment the decision actually happens. He mapped the whole idea out in a PowerPoint deck, then came to AppMakers USA to turn it into a real product, fully self-funded, as a first-time app founder.
"I had no investors. Everything was coming out of my savings. So I needed to make sure that my money was being well spent."
Every other app told him what he should have eaten. Jay wanted one that answered before the plate arrived.
We took the deck and built the app: an AI coach named Kai, food and menu photo analysis, glucose tracking, a personalized onboarding, notifications, and subscriptions, on iOS and Android. Every screen below is the shipped product.
Ask before you eat. Kai gives quick, actionable guidance, not a wall of text.
Readings tagged in range, high, or low, with a weekly average at a glance.
Sleep, habits, and history feed Kai so its guidance fits the person.
Add a reading in seconds, tagged by meal, bedtime, or context.
The full readings log, so patterns are easy to see over time.
Account, diabetes profile, and the subscription paywall we built in.
Screens show the app with illustrative sample data.
The same coach and glucose log, laid out properly for the bigger screen, plus a full cuisine-aware meal planner. Snap to Filipino, Mexican, Italian, or low-carb, and every meal comes with its macros and a glucose-friendly tag.
A week of glucose-friendly meals, filtered by cuisine, each with calories, carbs, and protein.
The weekly average and every recent reading, tagged in range, high, or low, on the tablet layout.
Snap a dish, a menu, or a nutrition label, and Kai reads it back: the macros, what to order, what to sub out. In testing it caught things a person would miss.
On Halloween, Jay pointed the camera at a bowl of candy while testing. Kai listed everything, including a couple of Nestle Crunch minis tucked so deep in the bowl you could not see them by eye. It had recognized a sliver of the blue-and-white wrapper.
"I looked in the bowl. It just saw the blue fold over and knew it was a Nestle Crunch. I was just like, how the hell did it see that?"
A health AI coach has to be accurate and fast without getting expensive at scale. So our engineers did not just reach for the default. We benchmarked several large language models on response quality, speed, reliability, developer experience, and cost.
OpenAI won on the metric that mattered most here: the most accurate, user-aligned answers, with the performance and scalability to iterate fast, and no extra infrastructure to babysit. We choose the model, we do not chase the logo.
Apple and Google flagged it as a health app: it needed clear language that it does not replace advice from a physician. We added exactly what they asked for. Apple rejected it again anyway. Our PM Dovrat and our lead engineer went back in, pointed the reviewer to the precise disclaimer areas we had changed, and asked them to look again. It was approved a couple of days later.
In a health product, trust breaks fast if the AI oversteps. So we drew tight boundaries early: what Kai can recommend, and what stays deterministic and rule-based. That is also what keeps it actionable instead of overwhelming. Quick, clear guidance beats a chatty essay when someone is standing at a restaurant table.
Have an app idea mapped out and no idea how to build it right?
Get a Free Project Estimate →Jay had never shipped an app, had no investors, and a hard budget line. The build had to be honest, communicative, and on time. That is the part he talks about most.
Daily on Slack with our engineer and PM, weekly reviews on every release, and the next phase planned on the same call. Questions got answered the same day.
A founder wants everything. We walked Jay through what mattered most now versus later, cut what could wait, and held his hard budget line while still shipping a strong app.
We scoped the hours against a real calendar, holidays included, and targeted a November-December finish. We hit it, and the app launched on both stores.
"The team walked me through, Jay, technically this is more important than that right now. They made sure we stayed within the budget, but also put an app out there that was really good. Timelines and deliverables was spot on. Quality of the app, spot on."
A first-time founder on why he chose us, what the build was like, and whether he would do it again. Worth the watch.
"The best thing I liked about the whole experience was the communication. It was amazing."
"From the design, to seeing the birth of the app, to the final in December, it was great."
"That's why I would recommend you, because you're legit, man."
GlucoCoachAI launched on the App Store and Google Play in January 2026, and is slowly growing as an indie, self-funded app.
App Store + Google Play, verified July 2026Real users are telling Jay it helps, especially newly diagnosed diabetics who felt lost, the exact person he built it for.
Founder-reported, from the video interviewA first-time, self-funded founder came out of the build calling the experience amazing and vouching for us on the record.
Video testimonial, aboveGlucoCoachAI That is exactly where Jay started. We take founders from a rough idea to a shipped product on iOS and Android, honestly, on budget, and we stay to keep building. The first project estimate is free.
