Every golf app says it has AI now. Open one up and it tells you to "aim for the center of the green." Groundbreaking. Thanks.
The honest truth is that most of these tools know exactly one thing about you: your handicap index. They take that one number, add it to par, and dress the result up as insight. A 12 handicap at a par 72 gets told he will probably shoot around 84. You did not need a model for that. You needed a calculator and a guess.
We wanted the opposite. A caddie that knows your game the way your buddy who has played 200 rounds with you knows it. Where you bleed strokes. Which holes wreck your card. Whether you are running hot or cold this month. And here is the part nobody else does: what to expect from you on a course you have never set foot on, by reasoning from courses like it and players like you.
This is how we built it, and why "we do not know yet" is sometimes the smartest thing software can say.
The Problem With "Probably An 84"
A score projection built on handicap-plus-par is a horoscope. It is technically about you, vaguely, and it is the same for everyone who shares your number.
But you are not everyone with a 12. Maybe you are a 12 who stripes it off the tee and three-putts like it is a personality trait. Maybe you are a 12 who scrambles like a magician but cannot carry a forced 200-yard tee shot to save your life. Two 12s, two completely different games, two completely different score distributions on the same course. The handicap flattens all of that into a single bland number.
So the first thing we built was an expectation band, not an expectation number. Before you tee off, caddie.fun gives you a realistic range for the day, with an 80 percent confidence band, built from your recent form and your actual history at that specific course. Not "you will shoot 84." More like "your day probably lands between 81 and 86, and here is why."
And then we did the thing that almost no consumer golf app is willing to do.
We Taught It To Say "We Do Not Know You Here Yet"
The first time you play a course, an honest system has a problem: it has no history of you there. The lazy move is to fake it. Fall back to handicap-plus-par, print a confident number, hope you do not notice it is hollow.
We refuse to do that. If the model has nothing real to say, it says so, and it widens the band on purpose with a note: "Wider range, we have not seen you here yet." The band tightens, round by round, as we actually learn your game on that course. Calibrated humility beats a confident guess every time.
The same honesty runs through Caddie IQ, a 0 to 100 read on how well we actually know your game, with tiers from "Learning" to "Locked In." On your first couple of rounds it tells you flat out: "Learning, 2 rounds in." It does not invent a 73 out of nowhere to look smart. A number you cannot trust is worse than no number at all.
The Part That Makes It Actually Intelligent
Here is where it stops being a stats page and starts being intelligence.
Saying "we do not know you here yet" is honest. But it is not the end of the story, because we know two things that are almost as good as your history at this exact course.
First: we know how you play courses like it. Your home track is a tight, tree-lined par 71 with four brutal par 3s? When you load up a different tight par 71 with a stack of long par 3s, your record on courses that look like that travels with you. We turned every course into a feature profile (par mix, length, slope, rating, the shape of its par 3s, 4s, and 5s) and measured the distance between them. Your tendencies transfer to similar layouts, weighted by how similar they really are. A near-twin course you have played ten times carries real weight. A loose match you played once barely nudges the read. That is the difference between a model and a guess.
Second: we know how players like you score there. Other golfers in your skill band and current form have played that course, and their scoring distribution is a genuine signal about what the place will ask of you. We pool that, privately and in aggregate, never anyone's name and never a sample so small it could point back at a person, and use it as a starting prior when your own history is thin.
Blend those two together and a first-timer at a brand new course gets a real, personal projection instead of a shrug. And the moment you start building your own history there, your direct record takes over and the borrowed signals quietly fade. The system converges on you. It uses the crowd only until it knows the individual.
The same engine powers Blow-Up Radar: your three highest-risk holes at this specific course, ranked, with how many strokes they typically cost you. Not a generic course guide. Your personal danger list. And on a course you have never played, it still works, because it knows you struggle on long par 3s and it can see that holes 4 and 16 here are exactly that.
It Learns Whether It Was Right
A projection you never check is just an opinion. So every pre-round prediction gets frozen before you tee off. We record what actually happened. And once a week the model refits itself against that ledger: how much to trust your handicap versus your course history versus your recent form, and how wide those confidence bands really need to be to hit their 80 percent mark.
This is the flywheel. Every round you play makes the next projection sharper, not just for you, but for the way the whole engine reasons about form, courses, and bands. It is data-derived, not hand-tuned by someone's gut feeling about golf. The longer you play, the more the thing turns into a caddie that has genuinely been on the bag for years.
Where It Shows Up
A model is worthless if it lives on one screen nobody opens. So we wired the same brain into everywhere a smart caddie would actually be useful.
The AI caddie talks like it knows you. Ask it anything and it already has your number one scoring leak, your Caddie IQ, and whether your current practice focus is paying off. Coach mode is not reading from a generic tip sheet. It knows that your par 3 game is costing you a stroke and a half a round, and it can tell you whether the work you have been putting in is moving the needle.
Discover finds courses you will actually like. Not "courses in your zip code." Courses whose feature profile matches the ones you score well and have fun on. The same similarity engine that powers a cold-start projection also powers a recommendation worth tapping.
Crew Fit reads the group. Starting a round with friends? It uses real skill ratings, not just raw handicaps, to tell you how well-matched the group is and who pairs competitively with whom. A tight match is more fun. We help you build one.
The stakes get smart. When you set up a Nassau, the app can suggest a press grounded in your live win probability, not a number pulled from a hat. "Press 5: about 62 percent to win and the EV favors it." Honest when it is confident, honest when it is not, and never a coin-flip dressed up as an edge.
The Cut
Here is the thing about a real moat. Anyone can clone a tip card in a weekend. What they cannot clone is the accumulated read on your game, your courses, and the thousands of rounds that taught the model how golfers like you actually score. That compounds. Every round makes it harder to copy and more annoying to leave behind.
We did not build "AI for golf" because it tested well in a tagline. We built a caddie that knows your game, tells you the truth about how well it knows it, and gets smarter every time you tee it up.
Go play a round. See what it already knows. Then play ten more and watch it get scary.