Most baby apps start from a reasonable idea: record enough and patterns will show up, and the patterns will help. Feeds, naps, diapers, weights, milestones, moods. Log it all and the app will tell you something.

We made the opposite bet. The person opening a baby app is exhausted, holding a baby with one hand, and already worried they're doing it wrong. That person doesn't need another job. They need one small, true thing a few times a day, and then to be left alone.

What it does

Six short tips a day. Two are waiting when you open the app. Four more arrive through the day, at nine, one, five and eight o'clock local time, each as a single notification. A tip is one thing to know or do, two or three sentences, chosen for your baby's exact age in weeks, written by a person, with a citation you can open and read. Made by people. Checked by pediatricians.

That's the product. There's an Explore tab for browsing and a search box that copes with one-handed 3am spelling, but the daily six is the point.

Why six, and not a feed

More would be noise. A feed exists to be scrolled, and scrolling is the last thing a new parent needs more of. Six is enough to be useful and few enough that each one actually gets read. We tried more, early on. It was worse. The tips stopped feeling chosen.

Why nothing to log

The evidence for logging as a daily habit, for a healthy baby, is thinner than the apps imply. What your pediatrician actually weighs is weight gain, wet diapers in the first days, and how your baby seems to you. None of that needs a timestamped entry at 2:40am.

Logging also changes how you watch your baby. In 2018 a study presented to the British Sociological Association interviewed twenty parents who had been heavy users of feeding and sleep tracking apps and then stopped. Its author, Dr. Kate Orton-Johnson of the University of Edinburgh, reported that the apps led to "guilt, pressure and social comparison," and that parents who quit described being "released from a techno anxiety about their activities as parents." The public reviews of the biggest apps in this category show the same split: people who love the predictions, and people who caught themselves fretting over an entry they forgot to make.

And data you never collect can't be lost, sold or subpoenaed. More on that in the privacy post.

So the app never asks you to record anything. It does notice which tips you open, save or search for, uses that to pick better ones, and tells you in plain language why it picked each one. But there's no streak, no count, no chart, and no way to fall behind.

Why weeks

A six-week-old and a ten-week-old aren't the same baby, and both are "two months." The clinicians who study infants count in weeks for the first year, and so do most parents, until somewhere around the first birthday. The app does too, and the age advances on its own, so there's nothing to update. There's a whole post on this.

Why the tips are allowed to say "this passes"

A lot of what frightens new parents is normal. The frightening part is not knowing that. Frequent night waking in the early months is, in the AAP's words, "developmentally appropriate." Evening crying peaks around six weeks and eases. The four-month sleep change is development, not damage.

So a good tip often says: this is common, here's roughly when it eases, and here's the line past which you should call your pediatrician. Our content gate checks for that last part. A tip that says "this is normal" without saying when it stops being normal gets flagged before a person ever sees it. Reassurance with no boundary is just a quiet way of being wrong.

Why the voices

The one place we let ourselves be a bit silly. Every tip can arrive in the voice you choose: Calm, Emojis, Sentient robot, Savage granny, Gandalf, or Random, which picks one per tip. Five voices, or a surprise.

The wording changes. The advice doesn't. The medical substance is identical in every voice, and a voice that added, dropped or exaggerated a recommendation would be a bug. Most people stay on Calm. Some stay on Granny. We're not here to judge.

Why no account

We never needed one. The app asks for your baby's age in weeks. If you let it, it uses a coarse location, rounded to roughly eleven kilometers, to know whether it's a hot day. That's the whole profile. No name, no email, no password, no birth date. We can't leak what we never collected, and we didn't want to be one more company holding a file on your child.

Where this came from

We had recently become parents ourselves and couldn't find anything well made to help two tired people through the weeks after a birth. The first version shipped in 2024 with a few hundred hand-written tips, then went quiet for a long stretch. In 2026 we rebuilt it from the ground up: a new library of roughly 2,860 tips, a content gate stricter than we're entirely comfortable admitting, and the two apps that are in review as this is written.

The gate cuts both ways. We've already retired eight tips, a few of them because we'd said the same thing better elsewhere. A slightly smaller library beats one tip we can't stand behind.

What it will not do

It won't diagnose anything. It won't replace your pediatrician; a tip will tell you to call one more often than you might expect. It won't tell you your baby is behind, because it doesn't know your baby, only their age. And it won't get busier. If some future change makes the app louder or more demanding, we got it wrong, and we'd like to be told.

The NHS puts the whole idea better than we can, in its advice to parents of a crying baby: "do not be ashamed to ask for help." An app can be one small form of that. It shouldn't be another thing to keep up with.

Coffee & Cradle is on iPhone and Android. The first week is free, and after that it costs less than a coffee a week, which felt like the right unit.

Ask a new parent how old their baby is and you'll get weeks, usually until somewhere near the first birthday. Pediatricians, midwives, and the people who write growth charts answer the same way. Most baby apps round you to a month anyway.

We went with weeks. Here's why, and what it changes about the advice you get.

Months are too big for the first year

Take two babies who are both "two months old." One is six weeks, the other is ten.

• At six weeks, crying is at its peak. The AAP describes daily fussing that "usually peaks at about 3 hours a day by 6 weeks old," then falls to one or two hours by three to four months.
• At ten weeks, the social smile is in, the head is steadier, and for most babies the worst of the evening crying is on its way out.
• At six weeks, feeding is still finding a rhythm. By ten weeks, many babies have stretched the gaps and some are doing a longer stretch at night. The Henderson study of sleep consolidation found the fastest change in self-regulated sleep between one and four months, which is the exact span a single "month 2" label covers.

Give both families the same two-month advice and you're early for one and late for the other.

Clinicians already count this way

• Corrected age. For babies born early, the AAP tells parents to start with "your baby's actual age in weeks (number of weeks since the date of birth)" and subtract the weeks of prematurity. The whole method runs in weeks because months are too coarse to correct.
• Growth charts. The World Health Organization publishes its weight-for-age standards for the earliest period as separate tables from "Birth to 13 weeks," in weekly steps, before switching to the longer-range charts.
• Checkups. In the UK, the NHS schedules a new-baby review "within 10 to 14 days of the birth," a heel-prick test at 5 to 8 days, and a full physical examination at 6 to 8 weeks. In the US, the AAP's schedule starts with a visit at 3 to 5 days, then one month, two months, four months, six months. The early visits are days and weeks apart because that's how fast things change.
• Milestones. The CDC's checklists are organized by month, but each one is a floor: a milestone is something "most children (75% or more) can do by a certain age." When the CDC and AAP revised those checklists in 2022, they moved from the 50th percentile to that 75 percent threshold so a missed milestone would actually mean something. A floor at two months tells you what to bring up at a checkup. It doesn't tell you what to do on a Wednesday at seven weeks.

What it changes about advice

Timing gets tighter. Advice about the umbilical stump belongs in weeks zero to four and nowhere else. A heads-up about the six-week growth spurt is useful at week five and noise at week nine. The reminder to stop swaddling matters from the moment a baby starts pushing to one side, which the AAP says can be as early as two months, and matters less once rolling is old news. Every tip in our library has a start week and an end week. A few span the whole two years; most don't.

Expectations get looser. Weeks also make it easier to say "babies vary." A month-based app that says "your 4-month-old should be rolling" has turned a range into a deadline without quite meaning to. A week-based range, say rolling somewhere between weeks 14 and 28, shows the spread. The windows we use to time tips are wide on purpose, drawn from mainstream AAP, CDC and NHS summaries, and none of them is a verdict on your baby. They decide which sourced tip you see a little earlier. That's all they do.

Preemies, and after the first birthday

If your baby came early, use corrected age for developmental expectations, as the AAP advises, for roughly the first two years. In Coffee & Cradle you enter whatever age you and your pediatrician actually use; if that's a corrected age, enter the corrected age. The app doesn't know or ask for a birth date, so it can't second-guess you.

After the first birthday, weeks get silly out loud ("she's sixty-three weeks") and most families switch to months. We keep counting in weeks under the hood all the way to 104, partly because the arithmetic is cleaner and partly because the windows for things like the twelve-month sleep wobble and first words are still narrower than a month. You don't have to update anything; the age moves on its own.

What weeks don't do

They don't make anyone's baby early or late, and they don't replace a pediatrician who knows your child. A general tip is still a general tip. If something about your baby's development worries you, the right move is the same at week six and week sixty: say it out loud at the next checkup, or call sooner. Pediatricians would much rather hear a question that turns out to be nothing.

The clearest case for timing is the 4-month sleep regression, where a heads-up two weeks early changes the whole experience. The first twelve weeks, mapped the same way, are in the fourth trimester.

Coffee & Cradle asks one thing about your baby, their age in weeks, and picks six short, sourced tips a day around it. The first week is free, and the age keeps itself current.

Open most baby apps for the first time and you get a form. Email, password, your name, your baby's name, your baby's date of birth. It feels normal because every app does it.

Coffee & Cradle asks how many weeks old your baby is. That's the first screen, and as far as who-you-are goes, it's also the last one. We've said "no account" a lot in our marketing, and it's a phrase that can hide things, so this post is the long version: what it means technically, what we do hold, and what we gave up to build it this way.

The technical part

There's no sign-up. When you open the app, it creates a random anonymous identifier using Firebase's anonymous authentication, which Google's own docs describe as a way to "create and use temporary anonymous accounts" for people "who haven't yet signed up to your app." The identifier is a string of characters. It isn't linked to an email, a phone number, a name, or any account you hold anywhere else.

Everything the app stores hangs off that string, and nothing it stores can be used to work out whose string it is.

What the app does hold

We'd rather over-disclose than let "privacy-first" stand in for a list. Straight from our privacy policy, this is all of it:

• Your baby's age in weeks. Never a birth date. The app advances the age by itself, so you type it once.
• Optional context, if you want to give it. Gender, temperament, and a few yes-or-no answers like first-time parent, whether your baby has an illness or disability, whether there are siblings. Each one exists to pick better tips. Skip any of them.
• Coarse location, if you allow it. We round your coordinates to roughly an eleven-kilometer grid before storing them. That's enough to know it's a hot day where you are, so a sun-safety tip can show up the right morning, and not enough to find a household. Say no and everything else still works.
• Time zone and notification settings, including a push token if you turn notifications on, so the four o'clock tip lands at four o'clock.
• What you do in the app. Searches you type, and which tips you open, save, like, or linger on. This is how the app learns what you care about; the "why this tip" panel shows you how it was used. It doesn't log keystrokes, only deliberate actions.
• Basic usage analytics. We use Mixpanel to record events like app opens and session length, along with device model and operating-system version, so we can tell whether the app is working. These are tied to a random identifier, not to you.

That last item is why the App Store privacy label for Coffee & Cradle won't say "Data Not Collected." Apple's definition is strict, and we think it should be: "collect" means "transmitting data off the device in a way that allows you and/or your third-party partners to access it for a period longer than what is necessary to service the transmitted request." Analytics counts. We'd rather say so than game a label.

One more thing we want to say plainly. When you search, your query text may be sent to an AI provider to understand it or, when nothing in the library matches, to draft a tip. It goes without your identifier or profile attached, under terms that don't allow it to be used for training.

What we never hold

No name. No email. No phone number. No date of birth. No contacts, no photos, no precise location, no advertising identifier. There's nothing to sign up for and no password to forget. We don't sell or rent data, we don't share it with data brokers, and there are no ads.

That tip is what coarse location is for. On a hot day it moves up the list. That's the whole use.

Why we were stubborn about it

Two things about this category bothered us.

Most baby apps aren't covered by medical privacy law. HIPAA covers doctors, hospitals and insurers, not the app on your phone. In April 2024 the Federal Trade Commission finalized changes to its Health Breach Notification Rule specifically "clarifying its applicability to health apps and other similar technologies not covered by HIPAA," and made clear that a breach includes "an unauthorized disclosure," not only a hack. A regulator doesn't spell that out unless the gap is real.

The neighbors have a poor record. When the Mozilla Foundation reviewed twenty-five period, pregnancy and fertility apps and devices in August 2022, eighteen of them got its "Privacy Not Included" warning label, and its lead researcher described the apps' policies on sharing data with law enforcement as "pretty vaguely outlined." Baby-care apps are the next shelf over, and what they hold is arguably worse: a child's name and exact birth date, which together make a permanent identifier that can't be changed later.

We didn't want to be the ones holding that file. You can't lose data you never had.

What it costs

Building without accounts gives up some things, and you should know which.

• There's no password reset, because there's no password. If you delete the app, the anonymous identifier is abandoned and nothing we hold can be connected back to you. That's the design working as intended, but it also means we can't restore a profile from an email address.
• Deletion is an email, not a button. We have no contact details for you, so you write to us and we work with you to find your anonymous record and erase it. The address is in the privacy policy.
• We can't show you a big personal history, because we don't keep one. The app remembers what you saved and what you searched. It has no timeline of your baby's life, and it never will.
• We can't tell you anything about your baby we didn't learn from their age. No app should. A tip is general information about a typical week, with a source. Your pediatrician knows your actual child.

The voices

The tone you pick for notifications (Calm, Emojis, Sentient robot, Savage granny, Gandalf, or Random) is stored with the rest of your preferences. It's honestly the most personal thing the app knows about you.

Five voices, or Random. The line underneath each is the same: breast milk or formula meets all of a young baby's fluid needs, even in hot weather, and there's no need for extra water or juice before solids start. If you have a question about your own baby's hydration, especially in a heatwave, ask your pediatrician.

Read the policy

It's short, it's in plain language, and it's at coffeeandcradle.com/privacy. If anything in it is unclear, tell us. We'll fix the policy rather than explain it.

The no-feed, no-log, no-streak side of the same decision is in why we built a baby app that does less.

Coffee & Cradle is free for the first week, on iPhone and Android, and the first screen asks one question.

Every baby app says its content is expert-reviewed. Ours says "Made by people. Checked by pediatricians." That's a lot of weight on seven words, so here's the work behind them.

This is the path from a blank line to your lock screen, and the ways a tip falls off it.

What a tip is allowed to be

The rules about shape come first. A tip is one small thing a parent can know or do today: a title of five words or fewer, then one to three short sentences. It's aimed at a window of the baby's age in weeks and at a time of day. It carries a source, and the source has to be a real, live, specific page on a short list of hosts: government and university domains, the NHS, the World Health Organization, and a named list of children's hospitals and professional bodies such as the AAP. A homepage doesn't count. A commercial site isn't allowed at all.

Some things a tip can't contain. A medication dose, any amount in milligrams or milliliters, is an automatic rejection. The right dose is your pediatrician's to give, by weight. Shaming language gets flagged. So does a tip that reassures ("this is normal," "don't worry") without saying when it stops being normal or when to call a doctor. We call that a missing boundary.

Step one: the automated checks

Every draft goes through the same automated checks first, whoever or whatever wrote it. Drafts come from our writers working to a written brief. We also use software to draft at volume against that same brief, and nothing a model writes reaches a parent without passing the checks below and a human eye.

The checks cover the schema, the controlled vocabulary of topics and categories, the title length, the age window, the source's host, the dose rule, the shaming rule and the reassurance rule. Near-duplicates of anything already in the library are rejected. Topic has to fit age: starting solids at two weeks old is rejected on sight. And every source URL gets fetched. A "not found" kills the tip. Fabricated links are the single most common failure in machine-drafted health content, so we don't skip this one.

The checks also read each tip's five public voices, not just the neutral text, because a parent on Savage granny reads the granny version. A dose that shows up in one voice and not the others is still a dose.

Step two: three readers, then three skeptics

Tips that get through go to a machine review in tiers: a paraphrase-duplicate check using text embeddings, an editor pass for voice and whether the tip is something you can actually act on, and an expert-style pass with three questions. Is it pediatrically safe. Does it fit the stated age. Does the cited page support this specific claim, not just the general topic.

Anything safety-critical, and anything in the health or feeding categories, then goes to an adversarial round. Three independent reviewers, each with a different brief (safety, age-fit, evidence), try to refute the tip. A majority against it kills it. So does one of them calling it unsafe. It's the most expensive step, so it only runs on the risky subset, roughly one tip in six.

Step three: people who know the subject

Machine review is good at duplicates, voice and obvious faults. It can't tell you whether a pediatric claim is right. For that, the whole active library gets built into one review page and handed to clinical reviewers, who read on their own and never see each other's verdicts. Two of them can reach opposite conclusions on the same tip without knowing it, which is the point of a second opinion. A person then resolves the disagreement.

A reviewer's verdict doesn't change anything by itself. A named person makes the edit, deliberately, and every edit re-runs the checks.

Step four: retirement

Tips don't last forever. When a reviewer or a source audit shows one is wrong, we retire it: inactive, with a written reason and, where one exists, a pointer to the tip that replaces it. Its ID is never reused, because old links point at it. We've retired nineteen so far, which is another way of saying we got nineteen wrong. A few of them, with the reasoning, because this is the part that earns the trust line:

• A tip that said swaddling reduces the risk of SIDS. It does not. The AAP's safe-sleep guide says plainly that "swaddling doesn't reduce the risk of SIDS." The tip also ran to sixteen weeks with no instruction to stop at the first sign of rolling, which the AAP says can be "as early as 2 months." Retired; replaced by tips that say the opposite.
• A tip recommending a weighted swaddle. The AAP: "Don't use weighted blankets, sleepers, swaddles or other weighted objects on or near your baby." NICHD's Safe to Sleep lists weighted items among the things to keep out of the sleep area. No safe rewrite existed. Retired.
• A tip that said to freeze a teething ring. The NHS: "Never put a teething ring in the freezer, as it could damage your baby's gums if it gets frozen." It directly contradicted four correctly sourced teething tips already in the library. Retired.
• A tip suggesting a steamy room for a cough. It appears, almost word for word, on the NHS croup page's list of things not to do: "do not put your child in a steamy room or get them to inhale steam." Scald risk, no benefit. Retired.
• A tip that called drawn-up legs after feeds "totally normal" and said to wait a day or two. That posture, with inconsolable crying, is how its own cited AAP page describes a condition that needs care right away. Retired and replaced with tips that keep the reassurance and add the red flag.

Several others went because a correct tip on the same theme already existed and the weaker one only added risk.

The voices don't get a pass

Each tip can arrive in Calm, Emojis, Sentient robot, Savage granny or Gandalf, or in Random, which picks one per tip. A voice can change the phrasing. It can't change what the tip says. One that adds, drops or exaggerates a recommendation is a bug, and the checks read all five. When a tip's wording is corrected, its five voices get thrown away and rewritten from the corrected text, so none of them goes on saying the old thing.

Underneath all five: swaddling may help some babies settle, but it doesn't reduce SIDS risk on its own; keep swaddled babies on their back on a firm, flat surface, and stop swaddling once they start trying to roll.

What this doesn't make us

It doesn't make a tip medical advice. A tip is general information about what's typical at a given age, with a source. Your pediatrician knows your baby. All we know is their age in weeks. When the two disagree, your pediatrician is right, and most tips say so in their last sentence.

It doesn't make us finished, either. The library is roughly 2,860 tips and we expect to retire more. If one looks wrong to you, open its source from the "why this tip" panel, and if the page doesn't back it up, tell us.

Why there are only six tips a day, and why none asks you to log anything, is in why we built a baby app that does less.

Coffee & Cradle sends six of these a day, chosen for your baby's age in weeks. The first week is free, and every tip carries the page it came from.

The baby's finally asleep on you, and the first thing you feel isn't relief. It's that you forgot to log the feed, and now the chart's wrong.

If you've had that moment, this is for you. It isn't a post about how tracking apps are bad. Plenty of families get real use out of them, and some pediatricians ask for a few days of a log for a reason. It's about what happens when the log stops helping and starts grading you, and what you can safely let go of.

This happens to a lot of people

In 2018, Dr. Kate Orton-Johnson at the University of Edinburgh interviewed twenty parents who'd been heavy users of feeding and sleep tracking apps and had since stopped. What she reported to the British Sociological Association was consistent: the apps led to "guilt, pressure and social comparison." One mother told her, "We're so bombarded with ideals of motherhood and it's hard enough without adding in guilt." The parents who'd quit described being "released from a techno anxiety about their activities as parents."

Read the reviews of the big apps in this category and you'll see two stories side by side. Five stars for the predictions that worked, and, a bit quieter, someone who found themselves obsessing over the data, or says it started to feel like homework. Both are true. Only you know which one you've got.

What the chart can't see

A tracker knows what you typed into it. It can't see your baby's hands go loose at the end of a feed, which the CDC lists as a fullness cue. It doesn't know that crying "is often a late sign of hunger" (CDC again), so the feed you logged as late may have been on time by the signs that count.

Compare that with what your pediatrician uses to tell whether a newborn is feeding well. The CDC's list: you can "see and hear your baby swallowing," wet diapers climb to around six a day by day four, and "Your baby is steadily gaining weight." A scale at the checkup and a rough diaper count in the first week. No interval chart.

Sleep is the same. The AAP's line on young babies is that "Frequent waking is developmentally appropriate." A sleep log full of wake-ups in a two-month-old is a picture of a normal baby. It only looks like a problem because it's a chart.

Milestones, too. The CDC's checklists describe what "most children (75% or more) can do by a certain age," which means one in four perfectly healthy babies will be past the listed age on any given item. The lists exist to start a conversation at a checkup, and that's about it.

Stepping back, a bit at a time

You don't have to delete anything today. Try these in order and stop wherever it feels right.

1. Ask your pediatrician what they actually want from you. Often the honest answer is "nothing, unless I ask." If there's a medical reason to log (slow weight gain, jaundice, a feeding plan for a premature baby), log that one thing for as long as they say, and nothing else.
2. Turn off the predictions and reminders. Keep the log if you like it. Lose the part that tells you when the next nap "should" be. That's a guess about an average baby, and nobody has one of those.
3. Stop logging whatever stresses you most. For most people that's feeds. Use the three signs above instead: swallowing, diapers, the scale.
4. Log for a week, then don't for a week. Notice which week you looked at your baby more.
5. Delete it, if the relief is bigger than the loss. What's in there isn't medical history. Your pediatrician has the weights.

Where to put the attention instead

On the baby, in a practical way. Hands at the mouth before the cry starts. Open palms after a feed. A head that lifts a little higher in tummy time this week than last. Every tracking app is a rough stand-in for these, and you're sitting right there.

Five voices, or Random, on the same line: offer about 2 to 3 ounces per feed in the early weeks, increase gradually, and let hunger and fullness cues guide the amount more than a set number.

When it's not really about the app

Sometimes the app isn't the cause, just where the worry has landed. If the worry is constant, if you can't sleep when the baby sleeps, if you're checking the baby's breathing many times a night, or if the low mood of the first two weeks hasn't lifted, please say so to a doctor, midwife or health visitor. The NHS is clear that postnatal depression is common, that "Fathers and partners can also have depression after having a baby," and that it's "important to get help even if you only have some of the signs."

In the US, Postpartum Support International runs a HelpLine at 1-800-944-4773 (call or text; it's not a crisis line). Your own doctor and your baby's pediatrician both want this conversation. Postpartum anxiety is treatable, and the treatment isn't a better chart.

What we did instead

Coffee & Cradle doesn't have a log, a chart, a streak or a count. We left them out on purpose, and the reasoning is in its own post. What it does have is six short tips a day for your baby's age in weeks, each with a source you can read, each saying what's normal and when to call. It notices which tips you open so it can pick better ones, and tells you why it picked each one. There's nothing in it you can fall behind on.

If you're wondering how a baby app works without ever asking who you are, that's the privacy post.

Coffee & Cradle is free for the first week, on iPhone and Android. There's nothing to enter, so there's nothing to forget.