Choose a series, then explore its episodes.
We work backwards from the chores you hate (the on-hold, the forms, the renewals) and show Private Agent One just doing them. Product preview; categories shown, not brand partnerships.
▶◆ Interactive~90 secondsOne dot is one hour. A year of them blooms, and the hours you spend following someone else's steps light up inside it - and honestly, it barely shows. Then we do it again for every year you work, and twenty-seven hours becomes 1,080: forty-five whole days, about six weeks. Silent, slow, and arithmetic you can check in your head. It ends the way the whole series does: you were never bad at this, you were simply too human for it - so here are your six weeks back.
The Hours · the centrepiece
▶◆ Interactive~15 seconds to loseA phone menu stands between you and a human. 🤫 Private Agent One is calling the same number. Press 1, press 4, press the one that quietly means please suffer — while One quietly gets through in eight seconds. You lose every time, and that is the entire joke: you were never bad at this, you were simply too human for it. Plays in about fifteen seconds, with hold music generated live in your browser.
Playable · Beat the Machine
▶◆ Interactive~30 seconds to your numberNot a film — something you play with. Tap the chores that are actually yours, and watch your year fill in: every gold dot is one hour you spend following someone else's steps. The arithmetic is shown in the open, the numbers are the ones you pick, and nothing is collected — it all runs in your browser. Then it hands you your number, and each chore links to the episode about it.
Interactive · The Chore Clock
▶⏵ Film10 shortsThe whole series, one tap. Open it and the first short plays; flick up for the next, tap for sound. Ten everyday chores most people hate, ten times 🤫 Private Agent One just does it - vertical, hook-first, about fifteen seconds each. The gesture you already know; zero clicks.
Not It. · The Feed
▶⏵ Film1:51Be honest: when was the last time you called customer service and enjoyed it? The menu that never has your option, the hold music, the robot swearing your call is important - for the fortieth minute. Fighting customer service was never hard; it's just tedious - a script, a hold, a form, a rule. And following the steps patiently, perfectly, a thousand times without getting bored is the one thing a great agent is built to do. Meet 🤫 Private Agent One: it makes the call, waits on hold cheerfully and endlessly, works only for you, and touches nothing without your handshake. It holds. You live. The stuff nobody wants to do is the first thing One takes off your plate.
Not It. · Ep.1
▶⏵ Film1:40From memory, add up every subscription you're paying for right now. You missed some, didn't you? It always starts innocently - a free trial, a cheerful cancel anytime - and then one little charge, times a dozen, quietly leaves your account every month while you're not looking. None is big enough to notice; together they're a slow leak in the bottom of the boat. You didn't overspend - you just forgot to cancel. Meet 🤫 Private Agent One: it reads every recurring charge, lines them up honestly, and cancels only what you say - only with your handshake. Keep everything you love, drop everything you forgot, and the money that was leaking away comes right back where it belongs. Yours.
Not It. · Ep.2
▶⏵ Film1:47It's nine p.m. You're finally home, dinner's done - and there it is, still waiting. The expense report, the one piece of work that followed you all the way home from the office. A month of little paper ghosts, and then the form: date, amount, and the dreaded select a category - a dropdown full of codes nobody remembers. But this isn't thinking work, it's copying. Every receipt just becomes a row. Meet 🤫 Private Agent One: it reads every receipt, matches each to the right line, picks the right code, and lines them all up - neat, totaled, quietly correct. You tap submit, only with your say-so. The chore that followed you home just went away - and your evening is yours again.
Not It. · Ep.3
▶⏵ Film1:33You bought the thing. The thing was wrong. And now, to get your own money back, you have to complete a small, joyless quest called the return: find the box, print the label, pick a reason from a dropdown that never quite fits, and haul it somewhere before a deadline you'll forget. Then you wait, while your money sits somewhere marked 'refund pending.' But it's a label, a box, and a drop-off - an errand, not brains. Meet 🤫 Private Agent One: it starts the return, picks the reason, makes the label, and books the pickup - then you approve with a tap, only with your say-so, and your money stops pending and comes home.
Not It. · Ep.4
▶⏵ Film1:31It's 2 a.m. Your flight just got canceled by a text message, and your reward is to call an airline and wait, wide awake, for a human. The hold music, the 'all our agents are busy,' the dread that the one good seat left is vanishing. But rebooking isn't genius work - find the next flight, hold the seat, confirm it. Meet 🤫 Private Agent One: it watches the flights, grabs the best seat the moment it opens, and lines up your new itinerary while you sleep. In the morning you glance and tap approve, only with your say-so. You're rebooked, and you never heard a single note of hold music.
Not It. · Ep.5
▶⏵ Film1:31You need one appointment - the doctor, the dentist, the car. Simple, right? Except booking it somehow takes three calls and a small miracle. You call; they're closed. They call back; you're in a meeting. You call again; you're on hold. This is phone tag, and nobody ever wins. But it's find-a-slot, make-the-call, confirm-the-time - coordination, not rocket science. Meet 🤫 Private Agent One: it calls, it waits, it books - the doctor Tuesday, the dentist Thursday, the garage Friday - and drops all three straight onto your calendar, only with your say-so. Three appointments booked, and your phone never once rang.
Not It. · Ep.6
▶⏵ Film1:31You're minding your business when your phone buzzes: a charge, a big one, for something you're almost certain you never bought. So begins the dispute - dig up the statement, explain yourself to a form, prove a negative to a company that would rather you gave up. There goes your afternoon. But disputing a charge is a recipe, not a riddle: gather the evidence, file the claim, follow up. Meet 🤫 Private Agent One: it pulls the records, files the dispute, attaches the proof, and chases the follow-ups - the whole tedious case, pressed while you get on with your day. You approve every step, only with your say-so, and the money that vanished gets reversed, right back to you.
Not It. · Ep.7
▶⏵ Film1:35Once a year, like clockwork, they come for you: the insurance, the registration, the membership, all whispering the same dreadful word - renew. None of it feels urgent, so you ignore it, right up until it's suddenly, stressfully overdue and you're overpaying out of panic. But a renewal is three moves: track the date, compare the price, click renew - a calendar and a checklist nobody's keeping. Meet 🤫 Private Agent One: it watches every date, shops each one for a better price, and handles the renewals before a deadline can sneak up - checking each with you first, only with your say-so. Covered for the year, and you never once felt the dread.
Not It. · Ep.8
▶⏵ Film1:39Name. Email. Address. Date of birth. You've typed these exact facts about yourself so many times you could do it in your sleep - and tonight you basically are. Same information, same order, into yet another slightly different box: the intake form, the signup, the checkout, again and again. You're not deciding anything; you've become a very expensive human photocopier. Because that's all this is - copying known facts into empty boxes. Meet 🤫 Private Agent One: it knows your details (held for you, by you) and drops them into every field instantly, then you review and submit - only with your say-so - and your information stays yours, shared only where you allow.
Not It. · Ep.9
▶⏵ Film1:38You just want to buy the thing for a fair price. So you open one tab, then another, then forty - and now you're a part-time, unpaid pricing analyst. The same item, a dozen prices, each with a coupon, a catch, or shipping that erases the discount. An hour later you're exhausted and still not sure someone, somewhere, isn't paying less. But finding the best price isn't taste - it's a search: check every seller, subtract the fine print, surface the lowest honest number. Meet 🤫 Private Agent One: it compares every seller, reads the fine print, and points to the genuinely lowest price - then buys at it, only with your say-so. The finale of the series: all the stuff nobody wants to do, handled by the one at your side.
Not It. · Ep.10 · Finale
Capable on day zero, and it compounds as you teach it. The interactive, plus a self-paced onboarding deck for every product.
▶◆ InteractivePlayAn interactive piece: 🤫 One is capable on day 0 with the subscriptions you already pay for, and compounds as you teach it, graduating with your consent to acting, paying, and collecting your whole life's documents.
Interactive
Six personas, the objections each one actually raises, and the honest answer to every single question — the working script for talking about Private Agent One without overselling it.
Sales training
How to explain a private supercomputer you own outright: what it does on day one, what it costs, who it is for, and where the honest limits are.
Sales training
The 🤫 Tag, explained the way it should be sold — a safety and wellness wearable, never a medical device, with every claim kept inside what it actually does.
Sales training
A supercomputer where the car sleeps, and idle hours earned back to the neighbourhood grid — the pitch, the economics, and the objections that come with both.
Sales training
A spare warehouse bay becomes token production: owner-first, consent-first, a receipt for every job. What to promise and what to label as roadmap.
Sales training
The warehouse-scale version of the same argument — capacity, consent, receipts — and how to size it honestly for a prospect.
Sales training
Watch the library back to back.
The engineering program: start with the program, then meet all nine machines. Open to everyone; hypothetical, not an offer.
▶◆ Interactive9 machinesThe set of great projects Hushh quant engineers crave to build. Markets are noisy because humans love drama - and an Alpha Seeking Machine is a specialized harvester of one flavor of that mispricing. The one idea (human emotion is a renewable, overpriced resource; we farm the gap), the shared contract that makes a machine great (harvests a named drama, composes net-delta-long, constitution-bound, human-approved, measured, reproducible, honest at the edges), and the fleet of nine machines to build - Fear Harvester, Euphoria Harvester, Patience Engine, Citadel, Regime Sentinel, Roster Gardener, Corporate-Events Machine, Ledger of Truth, and the Conductor. Concept-level only; fall in love with one, and go make some aloha.
Program
▶⏵ Film5:27It's the worst afternoon of the year - screens bleeding red, everyone sprinting for the exit - and one machine on our desk just woke up smiling. Meet the calm one: it gets paid, in cash, today, simply to promise it would buy a great business later at a price it already adores. Two endings, both wonderful - keep the cash, or own the business on sale. It builds a whole staircase of these promises, so a crash becomes a shopping list; it waits for real, sustained fear (measured, not felt); and it only ever promises what its cash and its love can both back. The louder the panic, the fatter its paycheck. Fear in, aloha out.
Meet the Machines · No.1
▶⏵ Film5:25The party's at its peak, the room is bone-certain a hot stock only goes up - and leaning against the wall, perfectly relaxed, is the cool one, quietly taking orders. Meet the machine that sells the crowd its certainty: on a name we already love, after it's run hot, it sells a slice of the top at a price we'd happily take, and pockets the premium. It trims the party; it never ends it - the large majority keeps all its upside. It waits for genuine froth (giddy, not merely green), sells only what it owns, and rolls forever. Peak confidence is peak payday. Euphoria in, aloha out.
Meet the Machines · No.2
▶⏵ Film5:43Two machines are pouring money onto the desk - a stream from fear, a stream from greed. So what do we do with all of it? Meet the quiet, relentless heart of the whole operation. It has one instinct: every dollar buys more of the twenty-seven greatest businesses on Earth. Its signature is the ratchet - ownership clicks one way only, up, and never back, so an investor wakes up owning a little more every single morning. Three flywheels (growth, dividends, aloha) compound together; it never bets against the family; and it's really building one thing - someone's tomorrow. The engine is simple. The patience to let it run is the whole art.
Meet the Machines · No.3
▶⏵ Film4:59The flood arrives - a once-a-decade crash, everyone underwater at once - and the only ones who can go shopping are the ones who kept an ark. Meet the Citadel: boring on purpose, heroic on schedule. A deep reserve of the safest money on Earth, always earning, ready to back the Fear Harvester's boldest promises and buy outright bargains - and a line it will never spend below. It's a floor, never a ceiling: small in crises, vast in bubbles - a fortress that breathes. One day at the bottom can make an entire decade. Its deepest gift: because of it, we are never the forced seller. We never have to panic.
Meet the Machines · No.4
▶⏵ Film5:27A thousand headlines a day, every one screaming act now - and one machine whose entire job is to say, almost always, a calm and confident no. Meet the lookout, the skeptic, the bouncer at the door. Its enemy is noise itself - the daily soap opera built to make you act. It reads slow, cool instruments (overbought, oversold, the price of fear), measured not felt, and takes the market's temperature without ever catching the fever. Nine hundred and ninety alarms out of a thousand: ignored. It waits for a signal to sustain before it nods - and when it does, that nod is the downbeat the whole fleet was waiting for. Restraint is the edge. It would rather miss a good wave than be caught in a bad one.
Meet the Machines · No.5
▶⏵ Film5:19Picture a garden - not of flowers, but of the twenty-seven greatest businesses on Earth. It does not tend itself. Meet the quiet curator: it loves the garden enough to be honest about any single plant. Its enemy is narrative rotation - yesterday's darling, today's pariah, the crowd re-rating great companies on pure mood. It ignores the story and reads the soil, judging on decades of durable cash, not this quarter's headline. Every seat must be earned - no tenure, only excellence. When a name must go, it never yanks; it winds down gracefully by selling calls into strength, while the newcomer eases in through the Fear Harvester's patient puts. The slowest machine in the fleet, and the surest.
Meet the Machines · No.6
▶⏵ Film5:39It's the big game - ninety thousand on their feet, every camera on the star - and backstage, one machine is making sure not a single detail is ever dropped. Meet the Corporate-Events Machine: it finds glory in the details, done perfectly. Its enemy is distraction - while everyone watches the show, the boring events (dividends, ex-dates, splits, assignments) get fumbled. It captures every dividend and hands it to the Patience Engine; never gives one away by accident; carries every split clean so the ratchet stays exact to the share; and the instant a slice is called away, it's already handled. A thousand tiny leaks sink a great ship - so it plugs every one. It never gets bored: the ten-thousandth event, as perfect as the first. No applause. Just correct.
Meet the Machines · No.7
▶⏵ Film5:53Eight brilliant machines, every one insisting it did great - but talk is cheap and money is real. So who checks? Meet the incorruptible one, the referee with instant replay, who trusts absolutely no one - not the harvesters, not even itself. Its enemy is self-deception, the only drama that happens inside the building. It rehearses the nightmares - a once-a-decade crash before coffee - across hundreds of imaginary days; then it never grades its own homework, re-deriving the whole truth from the raw ledger. Two witnesses, never one. Did the reserve ever dip below its floor? Did the ratchet ever slip a share? Not a vibe - a verdict. Red is a gift. It turns investing from confident opinion into checkable evidence. We don't ask for trust. We prove it.
Meet the Machines · No.8
▶⏵ Film6:02Nine virtuosos on a stage, each certain of their part - and with no one leading, that's not a symphony, it's noise. Meet the last machine: the maestro, the calm center that makes the other eight play as one. Its enemy is cacophony. Its instrument is the daily cycle: gather the day, let the Sentinel judge, let the harvesters draft, check every Iron Rule against a simulated future - and then stop, and turn to a human, because nothing becomes real until a person says yes. The machines propose; the human disposes. It guards the whole fleet, not machine by machine: always net-long the twenty-seven, citadel up, ratchet climbing. The song always outranks the soloist. And when it's perfect, you don't hear nine machines - you hear one song. That song is someone's growing security. Now - engineers - go build one.
Meet the Machines · No.9
Eleven 100, 200 and 300-level sessions, from capable and curious to your first pull request. Hypothetical, not an offer.
▶⏵ Film11:06The front door, for fresh grads. Session one of an eleven-part, 100/200/300-level on-ramp that takes you from capable-and-curious to your first pull request. The big picture: we're a permanent-capital, AI-first fund that owns the twenty-seven widest-moat businesses on Earth and gets paid by the crowd's emotion - a stream we call aloha, reinvested forever. The culture (we celebrate proofs, not features - your job is to break the machine), what a quant engineer actually does, and the whole ramp: 100 orientation, 200 the working math, 300 reading and running the machine. No finance PhD required - just aptitude and curiosity. Intuition first; the Master Class has the proofs.
Ramp-In · 100 · S1
▶⏵ Film9:04Session two, the foundation everything rests on. A share is not a ticker; it's a piece of a real business - a claim on all the cash it will ever produce, worth that cash discounted to today (intrinsic value). Price is what you pay; value is what you get, and the gap is where every dollar hides. Two forces move price - fundamentals (slow, real) and emotion (loud, fast) - so meet Mr. Market, the manic partner whose mood is your opportunity, and the margin of safety that lets you wait for the fat pitch. Plus the four marks of a moat and its shapes, why size isn't value, free cash flow and how businesses pay their owners, and the big one: volatility is not risk - permanent loss of capital is.
Ramp-In · 100 · S2
▶⏵ Film9:11Options, built from zero - no assumptions, no jargon. An option is a paid choice: the right, not the obligation, to act at a locked-in price before a deadline. Call to buy, put to sell. The cleanest model is insurance, and we choose to be the insurance company collecting the premiums. With a round-number example (sell the 90-strike put, keep the $3 or buy at $87 - no bad outcome), payoff pictures for buyer and seller, and covered calls sold into euphoria. Value comes from just two things - time and uncertainty - splits into intrinsic and time value, and moneyness names where the strike sits. The seller's edge is real (decay works for us; the crowd overpays for peace of mind), we sell covered or cash-secured, and we roll forever - that renewing stream is the aloha.
Ramp-In · 100 · S3
▶⏵ Film9:11The session that ties the 100 level together. The house wins by structure, not luck - a small edge played many times - and we run the same honest math: sell insurance the crowd overpays for, on the businesses we love, and let time and repetition do the work. The order of preference that is the opposite of the crowd's - sell a put, then a call, and only then buy - never overpaying for a great business. The aloha stream and forever rollover; the ratchet that turns one way so net shares and net ownership always rise; heads-we're-paid-tails-we-own-more; and three engines - growth, dividends, and aloha - turning one wheel. Honest about the risk: not free money - covered or cash-secured, only on names we love, with a deep reserve.
Ramp-In · 100 · S4
▶⏵ Film9:10The 200-level working math begins - friendlier than its reputation. The four ideas a quant thinks with: a price is a random walk (returns, and why we take logs); returns are bell-shaped while prices are log-normal (floored at zero, open above), with an honest note that real tails are fatter than the neat curve; volatility is the width of the bell - the typical daily move - and it grows with the square root of time (1% a day is about 16% a year); and expected value is the compass, outcome times odds, that turns the 90-put into a positive-by-design bet and the law of large numbers into the house. An average is a promise only if you survive to collect it - so we diversify and never bet the farm.
Ramp-In · 200 · S5
▶⏵ Film8:46The most famous formula in finance, made a friend - no proof, just meaning. An option's fair price is common sense in symbols: what you expect to get, minus what you expect to pay, each weighted by probability and discounted to today. Phi is just the bell's running total from Session 5; d1 and d2 are standardized moneyness - distance to the strike measured in wiggles - and the tiny gap between them is the whole time value. The five inputs (spot, strike, time, rate, volatility) with vol as the one belief among four facts, and the astonishing thing the formula leaves out: the stock's drift. We price the shake, not the trend - no fortune-telling required. The formula is a fair-value ruler we hold against the crowd's price.
Ramp-In · 200 · S6
▶⏵ Film8:35The Greeks, without fear - each one is just a slope. Delta: how much the option moves per dollar, the hedge ratio, and roughly the odds it finishes in the money - which is how we pick strikes. Gamma: delta's acceleration, the thing that bites fast near the strike. Theta: time decay - the seller's income, the aloha drip named, and why we roll in the 6-12 week window. Vega: the ear for fear - and we're short it, so we sell the panic and win as it normalizes. Assembled, they draw the desk's fingerprint: long delta, short vega, long theta - patience monetized - a durable posture, not a clever one. First-order truths, handled with the humility that keeps them from making us reckless.
Ramp-In · 200 · S7
▶⏵ Film8:55Closing the 200 level: the market's own opinion of volatility, and when to act. Implied volatility is the formula run backwards - solve for the sigma that matches the price - the market's fear made visible; when implied sits far above realized, the crowd is overpaying and that gap is the seller's meal. We compute our own IV (bisection, auditable) and rank it by percentile, so the machine surfaces the 27 by the price of fear each morning. Then signals: RSI as up-vs-down on a slow, ~27-week-memory weekly dial, with the desk's two bright lines - above 69 harvest, below 40 accumulate - and 'sustained' as the noise filter. All of it in service of the first rule of the desk: Do Nothing, until price, moment, and quality all agree. Restraint is the edge.
Ramp-In · 200 · S8
▶⏵ Film8:45The 300 level opens by opening the actual codebase - a map, not a maze. The engine is built as seven concentric rings, inner to outer, each reaching only inward: quant → signals → overlay → rules → portfolio → engine → learning (pure math at the core, the messy world at the edge). The daily heartbeat - bars → signals → tickets → approve → fills → rules → reports - with the human approve step in gold: the engine proposes, the GP disposes. Determinism (same inputs, same outputs, to the cent) and the two headline Iron Rules - the 22% citadel and the ratchet - checked against the simulated post-trade book before anything is real. A glass box, not a black one.
Ramp-In · 300 · S9
▶⏵ Film8:43How we earn the right to trust a number - the scientific method as a daily habit. Every run is reproducible byte-for-byte (a pure function, no hidden state - so any difference is a bug with a fingerprint). The tests are the documentation - a doc can drift, a test that runs on every change cannot - and golden files mean nothing changes silently. The simulator runs the real engine across deliberate worlds (calm, bull, crash, whipsaw) so we meet our worst day on purpose, safely. And the independent auditor never lets the engine grade its own homework: it re-derives every invariant from the raw ledger - two witnesses, never one - and fails loudly. Don't fool yourself, and never fool an investor: prove it, then say it.
Ramp-In · 300 · S10
▶⏵ Film8:47The capstone - and it makes everything real. Your first two weeks, hour by hour: clone and install (setup boring on purpose), run pytest and watch it go green - your first proof - then walk the rings inside-out, reading each module beside its test. The most instructive hour of week one: break a golden test on purpose and predict the diff before you look (chase the surprises - they're the edge of your understanding). Then your first PR: small and real, with the honest one-sentence description written before the code, a test that proves it, and the code left a little clearer than you found it. Closes on the oath - verify before you claim, keep the machine's promises, and above all, elevate the human. Welcome to the desk - now go make some aloha.
Ramp-In · 300 · S11
The 500/600-level Master Class deep dive, plus the overview films. Made for our talent, open to everyone. Hypothetical; not an offer.
▶⏵ Film12:50Part one of the 500/600-level deep dive: a seven-part, ~105-minute series that goes all the way down - building, running, and using the machine to manage permanent client capital. Part I is the intellectual foundation: why a permanent-capital, AI-first vehicle; Simons's method and Munger's quality, refused as tribes and fused on purpose; weight as projected free cash flow seven years out, not market cap; and why we never overpay - sell puts to be paid to wait, sell calls into euphoria, buy outright only third. We teach, then we test: you don't get to believe the machine, you get to break it.
Deep dive · Part I
▶⏵ Film12:08Part two of the deep dive: the full Black-Scholes derivation, on screen, from the ground up. Geometric Brownian motion, Itô's lemma from a Taylor expansion (with (dW)^2 = dt), the delta-hedged portfolio whose randomness cancels exactly, no-arbitrage and the PDE where the drift mu vanishes - so we never need a view on direction - then Girsanov, risk-neutral valuation, one Gaussian integral, and the closed form C = S·Phi(d1) - K·e^-rt·Phi(d2), which is exactly bs_price in the engine. Seven steps, no magic, every boundary checked. We derive it; we don't quote it.
Deep dive · Part II
▶⏵ Film11:03Part three: the Greeks, proved and verified. Differentiate the closed form for delta and three terms appear - but the star identity, proved from the ratio of the Gaussian densities, forces two to vanish, so delta = Phi(d1) exactly, checked against a finite difference to six figures. The same identity collapses vega, which then dies in the wings - exactly where our ladder lives - so Newton's method blows up there and we choose bisection: strictly monotone price, a unique root by the intermediate value theorem, below machine precision in a hundred steps. We solve our own implied vol, never the feed's. The desk, in three Greeks: long delta, short vega, long theta.
Deep dive · Part III
▶⏵ Film10:37Part four: the mathematics meets the market. We read emotion, not direction - Wilder's RSI is an EMA with effective span 2n-1 = 27, a 27-week memory on weekly closes, so the gates are structural and we read the last two prints, not one. DNSA: do nothing first. Then the two symmetric engines - sell-the-wall (a call ladder ITM/ATM/OTM on <=30% of a holding, delta-governed, >=70% keeps full upside, rolling forever at ~80% captured) and the synthetic long (cash-secured puts that pay us to wait or convert to ownership at our price). The ladder is a constrained integer program, and we refuse the raw optimizer's vertex to protect the fat tail. Every short option is marked at fair value, never intrinsic - the conscience line.
Deep dive · Part IV
▶⏵ Film10:49Part five: a strategy becomes a fund when it gains a constitution it cannot break. The constitution is code - rules as data, validated at startup, verifiable from a git diff. The daily cycle is one deterministic, idempotent pass that proposes to the GP, who disposes. The Iron Rules engine judges the simulated post-trade book and rejects a trade only if a check worsens - no-worsening, not no-violation - so a fund below its reserve floor can still sell the calls that refill it. The citadel: >=22% of NAV always in Treasuries, modeled as a controlled process, with a 90-day rebuild feasible iff the premium rate clears the deficit, and deficit-mode calls into RSI>69 to heal it. The ratchet: shares per unit are monotone; the machine does not cheat, it raises a ticket. The permanent high-water mark is fairness, made mechanical.
Deep dive · Part V
▶⏵ Film9:30Part six: how we earn the right to trust our own machine - a number is true not because we like it, but because it survived an attempt to break it. A deterministic wind tunnel where every price is a pure function of scenario, symbol, and date through SHA-256 hash-noise, so a 500-session run is reproducible byte for byte. The real engine driven through calm, a -35% crash, the recovery, and the chop, in a sandbox that never touches the live book. Then the adversary: an independent auditor that re-derives every invariant from the ledger, trusting none of the engine's own checks - two implementations cross-checked, and a broken invariant stops the line. The learning loop measures every trade against doing nothing and never rewrites the constitution - the defense against overfitting. And we publish our own open problems and mistakes. A method, not a secret.
Deep dive · Part VI
▶⏵ Film10:54The finale: what it is actually like to run this machine for the humans who wire us their life savings. It begins with a wire - 100% into Treasuries the day it clears, the citadel a loaded spring - then patient deployment (paid to wait, never paying up). The GP cockpit collapses the whole day into one page and one /approve: the machine proposes, the GP disposes. Aloha income is fuel; ownership is the point, so net shares and net delta only ever rise. Opt-in risk tiers (leverage into disruption; extreme-risk paired puts so you cannot be forced out of ownership) and quarterly roster replacement done by getting paid. The objective is max E[ln W] - the growth-optimal criterion, not Sharpe, which punishes the upside we exist to capture - which surfaces the open problem at the heart of the desk: harvest too much premium and you can lower long-run wealth. The oath, and: welcome to the desk, now go make it more honest.
Deep dive · Part VII
▶⏵ Film13:50The complete curriculum film every engineer, quant, and scientist watches on day one. The thesis, the full mathematics derived on screen (Black-Scholes from the SDE, delta proved and checked against the code, why we solve implied vol by bisection), the engine as constitution-as-code, the reserve and the ratchet, the scientific method at scale, and the part no one else teaches - how we actually work, your first two weeks. We teach, then we test: you don't get to believe the machine, you get to break it. Training, open to everyone. Every number on screen was checked against the code.
Onboarding film
▶⏵ Film10:08The keynote cut of the quant-engineering deep-dive: the mathematics of aloha derived on screen, module by module. Black-Scholes from the SDE, the delta-hedge that makes drift vanish, the identity that forces delta = Phi(d1) exactly, vega's collapse in the wings, and why we solve implied vol by bisection, not Newton. It closes on fair-value marking - the conscience line - and the permanent-capital objective. A tighter, faster companion to the full Onboarding. Every number on screen was checked against the code.
Keynote cut
▶⏵ Film7:52The derivation-driven seminar, built to be interrogated at depth. Black-Scholes from the SDE, the delta-cancellation identity proved so delta = Phi(d1) exactly, vega and why implied vol is solved by bisection not Newton, Wilder's RSI as an EMA with a 27-week memory, the fair-value marking that protects a permanent high-water mark, and the permanent-capital IRR - why not Sharpe, and when harvesting more premium lowers long-run wealth. Every number on screen was checked against the code.
Seminar
The open consent protocol every Private Agent is built on: the deep dive and the walkthrough.
▶⏵ Film20:12The 400-level deep dive: the six-phase handshake, the HCT token shipping today versus the CRT/DAT profile still proposed for v1, the scope grammar, the sealed envelope, the threat model, and the Apple Silicon reference implementation - benchmarked against OAuth. Every claim cited to the published spec, hosted by Nav and a skeptical systems engineer who won't take a claim without the wire format behind it.
Deep dive
▶▷ StoryReadThe end-to-end user story on four representative humans: the six-phase consent handshake, then the aha moment where One connects the dots your own mind keeps apart, strictly within what you granted.
Scrollable story
