The algorithm of a cry: How startups are turning baby whimpers into data

And so, they turned to a smart crib, made by Cradlewise, a company they already knew from the hardware world.

“It’s a fairly expensive purchase and we thought about it 20 times,” Mody said. The smart crib costs around ₹1.6 lakh. But eventually they decided it was best to be prepared and ordered it.

Within two days of his birth, he was sleeping in it. The early months, when most parents pace hallways, rocking babies in the dark, played out differently in their home. The crib’s motion picked up where their arms left off. A built-in webcam narrated the baby’s small routines and even a TV murmuring in the same room didn’t disturb him as the white noise in-built in the crib cancelled it out.

In Kolkata, Trisha Bhattacharjee, an advertising professional now 10 months postpartum, remembers the moment she stumbled upon Pukaar.ai, a parenting-tech startup. A week after delivery, as she was scrolling through Facebook during a break in the baby’s bawling, she saw an ad for Pukaar and downloaded the app. The app gave her a sliver of sanity.

“In the early days, every cry felt the same. The cry analyzer was helpful as I’d record my baby’s cry, upload it, and it would tell me if she was hungry or sleepy or in pain,” said Bhattacharjee.

Pukaar.ai has racked up more than 20,000 installs and 15,000 users in just six weeks since its launch, according to founder and chief executive officer (CEO) Karan Birpali.

For generations, parents of newborns have muddled their way through, relying on instinct, guesswork and advice. Now, a niche set of Indian startups, including Cradlewise, Pukaar.ai, Aignosis and Gabify, is treating early childhood as a data problem. Babies may not speak, but they communicate constantly. Every cry, stretch, blink and shuffle is a signal, and these young companies believe that decoding those signals can give parents some clarity in the most uncertain months.

Pursuing their individual paths, these babytechs are finding takers across India, which is projected to be the world’s most populous country at least till the end of this century. For the startups, that is good news.

Globally, especially in the US, baby tech has moved far ahead with smart monitors, bassinets (small, portable beds) and analytical tools, which have been in the market for years. India, by comparison, is at the very beginning. Beyond basic consumer products, there has been little tech-led innovation in baby care.

Investors are beginning to take note, with early capital flowing into the space on the belief that AI-led infant care could become a large consumer and healthcare category.

“These are not the large language models people see in everyday AI (artificial intelligence) products. Our system uses signal-processing models that read frequencies rather than text,” said Birpali. “A baby’s cry has no words, only patterns of sound, so we analyse its frequency, amplitude and breathing gaps to create a visual map known as a spectrogram. That allows the model to distinguish hunger from discomfort, pain or a cry for attention.”

Cradlewise, which was founded in 2016 but launched in India just last year, takes a motion-first approach. A camera sits on an arc above the smart crib, supported by sensors that track how a baby is sleeping. When a child begins to stir, the crib reads those early cues and starts a gentle bounce with soothing sounds, stopping once the baby has settled again. Over time, the system learns the infant’s patterns.

The crib begins with what it has learned from all Cradlewise users and then adapts itself to the behaviour of each child, said Srishti Mahajan, head of product at the company.

Aignosis and Gabify, both founded in 2024, look at development rather than day-to-day cues. They use standard webcams and AI models to flag early signs of conditions such as autism, attention-deficit/hyperactivity disorder (ADHD) or speech delays. Parents upload short videos of their child, and the systems analyse eye movement, attention, facial cues and other behavioural markers.

While Aignosis uses eye-tracking, Gabify reviews videos across 189 parameters, compares its findings with clinical frameworks, and then generates a screening report.

India’s parenting market is valued at roughly $26–28 billion, covering everything from nutrition to edtech to wearables. “The market is not the question,” said Ankur Mittal of Inflection Point Ventures. “The real question is whether AI applied to infant behaviour becomes a platform, not just a product.”

According to Mittal, a cry-detection tool alone cannot support a large business. “If you’re building a cry-detection app, you’re a feature. If you’re building a developmental intelligence layer that follows a child from birth to age five and gives parents, paediatricians and schools a shared language for early milestones, then you’re creating a category.”

With 25 million births a year in India, even a small penetration among top income groups creates a significant base. “But founders have to earn the right to expand. You can’t claim a $5 billion market on day one if you’re selling a smart bassinet.”

The companies Mint spoke with are now aiming to build an integrated ecosystem as the real value lies in the data. Continuous, structured behavioural data from infancy is rare in India, yet it can be highly valuable for pharma companies working on early-intervention drugs, insurers modelling long-term child health, and governments shaping public-health policy. Mittal notes that whoever builds the first large, longitudinal dataset, which is clean, anonymized and linked to developmental outcomes, will have a lasting advantage.

Cradlewise is already working in that direction. The company has years of sleep data from thousands of babies and it is using this data to build a new product, Sleep Coach, a conversational AI tool. “It’s essentially a software product that sits on top of your baby’s data and then gives you suggestions on how to improve the sleep,” said Mahajan. The tool is still in early beta and will eventually be folded into a subscription layer.

Gabify is building a different kind of platform. Its model brings therapists, developmental paediatricians, teachers and parents onto a single virtual SaaS system so that everyone involved in a child’s care can track progress and follow a unified plan. If a child is undergoing therapy, each stakeholder, be it a clinic, school, or parent, has access to shared assessments, recommended activities and milestone tracking.

Pukaar.ai, too, is taking a full-stack approach. The company began with cry analysis but soon realized parents needed broader support. “Eighty percent of parents don’t fully trust the decisions they’re taking,” said founder Birpali. Qara, the AI companion Pukar built, remembers daily events, connects patterns, raises red flags and elevates anything that needs medical attention. Pukaar has also onboarded doctors who review weekly baby-health summaries generated by Qara and handle consultations.

Pukaar.ai says its cry detection model is trained on hospital-labelled datasets. The system first identifies whether a sound is actually a cry, with an accuracy of about 92-93%. It then moves to classification, where it is close to 99% accurate in distinguishing hunger from non-hunger cries, before going deeper into categories such as discomfort or pain, according to Birpali. Doctors regularly test the system and suggest improvements, and the model is updated continuously based on feedback, he told Mint.

Cradlewise looks at accuracy differently. Instead of labels, it tracks outcomes. “If the crib can detect the right moment and prevent a baby from waking up, that’s our true metric,” said Mahajan. “On an average, we see that babies have about two hours of extra sleep in a day,” she said.

For startups working on developmental screening, the benchmarks are closer to clinical standards. Gabify says its model shows around 90% accuracy based on internal studies conducted with preschools, therapy centres and doctors. External validation from bodies such as AIIMS and ICMR is still in progress.

Aignosis, which uses eye tracking to flag early signs of autism, reports accuracy in the high 80s for both sensitivity and specificity. In simple terms, that means it can correctly distinguish between children with and without signs of the condition in most cases, though results can vary depending on how the test is conducted. Aignosis and Gabify, however, are careful to stress that their platforms are screening tools, not diagnostic ones.

But consent and ownership is a big challenge. “In India, the regulatory framework around child data is still evolving under DPDP (Digital Personal Data Protection Act). The parent must own the data and any startup that tries to quietly monetize infant data without explicit, informed consent is sitting on a reputational time bomb,” said Mittal.

Cradlewise’s Mahajan explained that the crib and the parent’s phone usually run on the same Wi-Fi network. In that setup, the live video stream stays within the home network. “Your video feed isn’t sampled or sent to our servers,” she said.

Parents can also choose whether they want to allow data uploads for model improvement. Switching off uploads keeps everything on-device, but the crib then relies only on a general alert system instead of personalized insights. Enabling uploads means the crib occasionally samples short snippets from sensors or video when something changes, encrypts it, and stores it on a secure server.

Other startups in the space follow a similar approach. Gabify’s co-founder Sahil Chopra said the company built its own encryption software so that any data being uploaded is protected before it reaches the cloud. The system also auto-deletes everything after seven days.

Pukaar.AI founder Birpali gave a similar assurance for the company’s infant-care assistant. “Qara keeps all your information private. Everything is encrypted,” he said.

Many clinicians say these systems are most useful for spotting early deviations in behaviour, such as unusual crying intensity, irregular sleep, or developmental inconsistencies. They can also reassure first-time parents who are overwhelmed by ambiguity. But experts warn that over-reliance can chip away at a parent’s confidence over time.

“Early caregiving is a critical period for responsive interaction,” said Dr Roomki Mitra, a consultant child psychologist at Kailash hospital in Noida. She explained that emotional security is built through small, consistent exchanges: the way a parent notices tone, timing, eye contact and body cues. “If parents start depending more on external prompts than their own reading of the child, it can subtly affect the rhythm of bonding.”

Doctors echo this caution. Technology can offer awareness, but it cannot replace the sensitivity needed to understand a baby moment by moment, said Dr Preetha Joshi, a neonatal, paediatric and cardiac intensivist at Kokilaben Dhirubhai Ambani hospital in Mumbai. In rare but serious conditions, such as abdominal obstruction or signs of congenital heart disease, a cry mislabelled as a minor discomfort could delay crucial treatment, she said.

Some acknowledge frequent edge cases, as in the case of Cradlewise, when babies wake up quietly or when they stand or jump in cribs and need immediate attention. “These are limitations we are working around,” said Mahajan.

Investors point to deeper challenges as well. Trust is fragile in infant care. A false alarm at 2am can create real panic. Retention is another hurdle, since babies outgrow products quickly. And for hardware-heavy startups, margins are tight unless parents continue paying for software services. “Adoption and pricing will be a real challenge. Many parents will hesitate if it feels risky or too expensive,” said Anil Joshi of Unicorn India Ventures.

In the end, the real moat may have little to do with algorithms. “The real defensibility comes from behaviour change. If a mother opens your app before she calls the paediatrician, you have won,” said IPV’s Mittal. In the babytech race, that is perceived as a sign of success.