We're building the product
that "would have been a lifesaver back then"
Data-driven U.S. admissions guidance · no made-up numbers · no promises · every judgment auditable
I'm a Chinese-American kid myself — UCSD class of '19.
I know the anxiety when parents stare at a single admit-rate number.
I tried using ChatGPT to help an applicant in my circle build a school list. The "32% admit probability" it gave was made up out of thin air: training data is a snapshot from a year ago, it doesn't know the latest admissions policy changes, and it has no read on the real applicant density in Chinese-American hubs.
Parents prep for four years based on this kind of "probability" — and only realize at the deadline that the direction was wrong. That's why I'm building PeiPaoLab — not because admissions is a great business, but because this shouldn't come down to luck.
So we built the Kaiso engine — recent 8 admission cycles + current-season admissions policy, refreshed every cycle, with the reasoning behind every judgment laid out for the family. We don't make up numbers, don't promise outcomes — only three-tier classification + the reasoning that backs it.
Four numbers hold up the engine
hardware / soft strength / policy / identity / geography — multi-dim crossing
8 recent admission cycles, covering a full policy rotation
Admissions policy refreshed every cycle, keeping pace with annual changes
Real competitiveness density mapped across Chinese-American hubs
Four iron rules
As long as PeiPaoLab is running, these four things will never happen. Not "not yet" — never.
Never make up admit probabilities
When a generic LLM says "32-45% probability", it's making up numbers. We only do three-tier classification (Reach / Match / Safety) + the reasoning behind it.
Never promise admit outcomes
Anyone claiming "guaranteed admission" is lying to you. We only give data and direction — no promises.
Never offer 1-on-1 consults / essay ghostwriting / application services
We make tools, not services — the data, algorithms, model suggestions, and content are for the applicant and family to use directly.
Never sell user data to third parties
Applicant grades and activities are sensitive. We only use them to serve your own family.
Common questions
DIFFERENTIATION
Q · How are you different from ChatGPT / Claude?+
Generic LLMs run on training-time data, so they tend to make up admit probabilities ("you have a 32-45% chance") and miss current-season admissions changes. Kaiso runs on 8 recent admission cycles + 13-dim scoring + current-season admissions policy, refreshed every cycle. We don't make up numbers — only three-tier classification (Reach/Match/Safety) + the reasoning behind it.
DATA SOURCES
Q · Where does your data come from?+
Public data: each school's Common Data Set, College Scorecard, IPEDS, official School Profiles. Chinese-American admit data: aggregated from public shares on 1point3acres, Xiaohongshu, and similar communities. Every source is publicly auditable.
Q · Is applicant and family info safe?+
Encrypted storage, used only to generate your own diagnosis and reports. Never used for ads, never sold to third parties. See privacy policy.
PRODUCT
Q · Applicant is still in elementary / middle school — does this work for me?+
Yes. Growth Archive is best built starting from late elementary or middle school — the earlier you start, the cheaper to course-correct. Subscription early-bird is free at the moment; leaving an email locks $9/mo lifetime — much cheaper than realizing in ten years that the path was wrong.
Q · Is the Sprint Pack really that cheap?+
In the early phase, yes. Beta $49, early-bird $99, post-launch $199. The report includes school list + essay direction + timeline + ED/EA strategy + EC playbook. Permanently archived in your account for the whole family to revisit, no hidden fees, no sales calls.
Q · You really don't ghostwrite essays?+
Really. Ghostwriting carries academic-integrity risk, and AOs are getting better at detecting it — ghostwriting is now the bigger risk. We only tell the applicant: given your raw material, here are directions you could take, what each is missing, and which angles are over-used. The essay the applicant writes themselves is the application that's actually theirs.
Q · Can I cancel the subscription anytime?+
Yes. One click in account center — no charge next month. Historical data retained for 6 months, then auto-cleaned per privacy policy.
Four data pools — each with verifiable sources
Admit case pool
8 recent admission cycles of real admit data
Admissions policy
Each school's official + Common Data Set
School profile
IPEDS / College Scorecard / School Profile
EC library
212 verified picks + soft blacklist
Data-driven · all the way
Pick the applicant's current stage to start ↓