Huckleberry taught a machine to know when your baby will sleep.

It's 2:14 a.m. somewhere, and a parent is staring at a phone instead of a crib. The app on screen isn't telling them to relax. It's telling them, with some confidence, that the next good nap window opens at 9:40. That parent is one of more than five million using Huckleberry - and that 9:40 is not a guess. It's a prediction, drawn from five billion logged moments of other people's exhaustion.

A sleep company that quietly became a data company

Huckleberry Labs sells calm. What it actually runs on is math. The Irvine, California company has built the No. 1 baby tracker in several countries by doing something deceptively boring: logging when children sleep, eat, and fuss, then turning billions of those entries into specific, timed advice. The friendly name and the soft branding hide a machine-learning engine that has been fed more infant sleep data than arguably anyone in history.

The product looks like a tracking app. The business is closer to a behavioral-medicine lab that happens to fit in a diaper bag. Parents open it for one reason - more sleep - and stay because it keeps being right.

Everyone has advice. Almost none of it fits your kid.

Infant sleep is an industry of opinions. Books, forums, grandparents, and a thousand blog posts all promise the secret - and they contradict each other by lunch. The cruel part is that real sleep consultants, the people who can actually read one specific child, cost more than most families can spend. Expertise existed. It just sat behind a price tag.

Then there's the timing problem, which is the one nobody warns you about. Put a baby down ten minutes too early and they fight it. Ten minutes too late and they're overtired and furious. The "sweet spot" is real, narrow, and moves every single day as the child grows. Parents were being asked to hit a target they couldn't see.

Two engineers, one baby who would not sleep

Jessica Toh was working at a software startup when she had her first child. The baby woke every few hours for twenty months. She and her husband, Seng Oon Toh, read everything and fixed nothing. Both are engineers - she has a background spanning electrical engineering, computer science, and business; he came from electrical engineering and R&D - so they did the engineer thing. They started logging data and looking for the pattern.

The bet underneath Huckleberry, launched in 2017, was simple and slightly audacious: that a child's optimal sleep windows are predictable if you have enough data, and that pediatric expertise could be encoded into software instead of billed by the hour. If both were true, you could hand a $200-an-hour consultant's instincts to anyone with a phone. They were betting expertise could scale.

SweetSpot, and a berry that already knows your kid

The centerpiece is SweetSpot - a real-time predictive algorithm that tells you, today, when this child is most likely to fall asleep easily. It learns from the child's own history, so the answer on a Tuesday is not the answer from three weeks ago. Around it sits flexible tracking for sleep, feeding, diapers, and growth, plus expert-built sleep plans that adapt instead of preaching.

In February 2026 the company shipped Berry, a context-aware AI that does the one thing a 3 a.m. Google search never could: it already knows your child's age, schedule, and last logged meltdown. No re-explaining. Parents in testing said it felt like talking to a pediatrician who'd read the whole file - because, in a sense, it had.

Five billion data points don't lie down quietly

Skepticism is fair - parenting apps over-promise constantly. So here are the numbers Huckleberry actually stands on. More than five million families. A 4.9-star App Store rating across 40,000-plus reviews. The No. 1 spot in the iOS medical category in multiple countries. And, the stat that matters most to a desperate parent: 93% report their child's sleep improved.

Make the expensive part of parenting free-ish

Huckleberry's stated aim is to make behavioral and pediatric expertise accessible and affordable to every family - not just the ones who can hire a consultant. Sleep is the wedge, not the whole plan. The company frames family wellbeing as a row of dominoes, where steadier sleep tips into steadier everything: calmer mornings, more patient parents, better days for the kid. The mission is less "sell an app" and more "stop letting good advice be a luxury good."

The 2 a.m. problem isn't going anywhere

Children will keep being born, and they will keep refusing to sleep on schedule. What's changing is that the accumulated wisdom of millions of nights can now be pointed at one specific, fussy, particular child in real time. Berry is the bet that the next decade of parenting help won't come from a search bar or a $300 consult, but from software that already knows the file. Whether competitors like Napper or Nanit can match a five-billion-point head start is the open question.

Back to 2:14 a.m. The crib is finally quiet. The parent didn't outlast the baby through willpower - they hit the window the app called at 9:40, then again the next night, and the night after, until the guessing stopped. That's the whole product, really. Huckleberry didn't promise a perfect sleeper. It just took the one thing tired humans are worst at - timing - and did the math for them.