Stories
The people we work for.
Feint exists for particular people in particular moments. A grandmother with the phone to her ear. A founder three time zones from home. A family at a kitchen table, deciding on a word.
The five people here are illustrative composites. None is a real client, and we do not publish real clients. We built each one from the patterns we meet in this work, and from the figures linked beside their stories, which are real and dated.
The habits are real. The work is real. Each story shows one thing we did, and a plain picture of how.
Ruth asked the question.
A retired schoolteacher in a kitchen full of plants. She practiced the worst call of her life, months before it ever came.
Ruth taught fourth grade for thirty-one years. She raised two children and now has four grandchildren. She keeps a landline because she likes the weight of the handset. She is not naive. She reads the paper, and she has heard about the calls.
Her son signed her up for Riposte after a neighbor lost forty thousand dollars. The neighbor had a grandson's voice in her ear, crying, and a lawyer who needed bail by noon. The voice was not her grandson. By the time the family understood, the money was gone.
Riposte begins with a conversation before any test. Ruth's lead met the whole family at her table. Together they agreed on a simple habit. If anyone calls in trouble and asks for money or codes, Ruth hangs up. She calls that person back on the number she already has. And before any of it, she asks a question only her family can answer. They chose their word together. In this house the word is the olive tree.
The ground
Among older adults, impersonation losses over $100,000 rose eightfold from 2020 to 2024, from $55 million to $445 million. A phone call was the first contact in 41% of those largest cases.
FTC Data Spotlight, “False alarm, real scam”, 2025-08. Context: adults 60+ reported $7.75 billion in losses to the FBI in 2025, up 59%. HousingWire, citing FBI IC3, 2026-04-13.
Then, months before any real call, Ruth practiced. This is the part people find strange until they try it. With her written consent, and a stop word she could say at any moment, Feint ran a bout. A voice called, upset, in a hurry, asking for help. It was a rehearsal, designed and supervised by a person on Ruth's side. Ruth felt the pull of it. Her chest tightened. Then she did the thing she had practiced. She asked the question. The exercise ended, and her lead walked her through it over tea.
“I knew the voice. I asked the question anyway. That is the part you have to practice.” Ruth, 78 — Riposte member
Autumn came, and the call came with it. The same shape, a familiar voice, the same hurry. Ruth's chest tightened in the same place. Because she had done it before, her hand knew what to do. She asked the question. There was a pause. Then the line went quiet, and clicked dead.
That is the whole story, and the quiet is the point. A practiced habit held under pressure. No money moved. Ruth made herself a cup of tea and called her son on the number she already had. He was fine. He had never called.
Ruth holds Riposte, $5,000 a month: the household tier, built around a family's verification habit and one practiced exercise a quarter.
The family that wrote a word down.
Three generations, one kitchen table, and a habit they can all remember.
The Raman-Castillos are four people and one rule. There is a grandmother, two parents, and a teenager who is faster with a phone than any of them. They came to Feint not because something happened, but because they did not want it to.
Most families already know the advice. Use different passwords. Turn on two-factor sign-in. Do not trust a number just because it looks right. Knowing is the easy part. The hard part is doing it, every day, when you are tired and the message feels urgent.
The ground
Across five years, regular use of two-factor sign-in fell from 94% to 53%. The share of people who always check messages for signs of a scam fell from 51% to 36%. People learn the habit, then let it slip.
National Cybersecurity Alliance, five-year “Oh Behave” findings, 2026-02-25.
Riposte starts every household the same way. Before any exercise, before any testing, the family designs its own verification protocol. The lead asks plain questions. How does money move in this house? Who is allowed to approve it? If Grandma gets a call from a grandchild in trouble, what does she do first? The answers become a short set of habits the family can actually keep.
The Raman-Castillos chose three. Money never moves on a single phone call. A request that creates a rush is a reason to slow down. And any member of the family can ask any other for the family word, at any time. The teenager suggested writing the word on a card and keeping it where everyone could find it. So one evening they sat at the table, and the youngest wrote it down.
“We did not learn anything we hadn't heard before. We just finally did it, together, and wrote it down.” The Raman-Castillos — Riposte members
The word itself does not matter much. What matters is that it is shared, that it lives nowhere online, and that everyone knows when to ask for it. A good family word is easy to remember and impossible to guess from a social profile. The family practiced asking for it until it stopped feeling rude between people who love each other. That is the real work. Verification has to feel normal, so that asking is never an accusation.
The Raman-Castillos hold Riposte, $5,000 a month, covering a whole household of up to six adults.
Priya shipped in a weekend.
A bookings app built fast, four thousand real customers, and no security team but ours.
Priya built her booking app over a long weekend. She is a designer by training and a builder by instinct. She described what she wanted to an AI coding tool, and by Sunday night she had a working product. By the next quarter she had four thousand customers. Their names, their phone numbers, and their appointment histories all lived in a database she had never thought hard about.
This is now an ordinary story. The people who build apps this way are not careless. They are fast, and they are alone. They own real users' data, and they have no security team.
The ground
In one twelve-month stretch, four separate scans found thousands of AI-built apps leaking personal data. One September 2026 study found about 16,000 exposed databases on a single hosting platform, holding names, addresses, and phone numbers.
TechCrunch on UpGuard's findings, “Some Supabase customers are publicly exposing reams of people's data”, 2026-09-25. A separate scan of 5,600 vibe-coded apps found 175 cases of exposed personal data. Escape, 2025-10.
Priya joined Parry for $499 a month. Parry is built for the security-curious founder and engineer. The center of it is the fleet: a set of agents that runs every night, on local models, checking the surfaces a builder actually has. Priya connected read-only access to her app and her accounts. From then on, the fleet kept an inventory and watched for drift — a new setting, a new deploy, a key left somewhere it should not be. Each night it ran a check and carried its findings forward.
For twenty-nine nights, the check came back clean. Priya barely noticed it, which is how it is supposed to feel. Then on one night, after a routine change she had made to ship a new feature, the fleet flagged something real. A database rule that was meant to keep customer records private had been left open. Anyone who knew where to look could have read them.
“I didn't ship a mistake on purpose. I shipped it at midnight, tired, like everyone does. Something was awake.” Priya, 29 — Parry member
No raw alarm reached Priya. An analyst reviewed the finding first, which is the discipline at every tier. Clients see confirmed, needs validation, or rejected, never the agents' raw output. By morning Priya had a plain-language note: here is what changed, here is what it exposed, here is the one setting to fix. She fixed it before her first coffee. No customer data ever left. The exposure existed for less than a day.
Priya holds Parry, $499 a month: nightly automated audits, with a human analyst who decides what matters.
The target was never Marcus.
A founder who travels every month, and the inbox two desks away that holds his whole life.
Marcus runs a company that is doing well enough to make him worth studying. He is on a plane most weeks. He has a good phone, a strong password manager, and the kind of caution that comes from reading the news. If you wanted to reach Marcus, you would not start with Marcus.
You would start with his assistant.
This is what the Master tier made visible. Master, at $30,000 a month, is for a person who is a target by position. It opens with a mapping exercise. Feint builds a careful inventory of the principal's real attack surface: devices, accounts, domains, the home network, and the office systems the office authorizes us to test. Then it asks a quiet question. Where does exposure actually concentrate?
For Marcus, the answer was not his own devices. It was an inbox two desks away. His assistant's email held his calendar, his travel, his approvals, and years of forwarded threads. It knew when he would be in the air and unreachable. It knew which vendors he paid, and how. In plain terms, it was the most exposed node in his life, and it did not belong to him.
The ground
62% of organizations reported a deepfake attempt in the past year, often an impersonated executive used to trigger a transfer. One firm lost $25.6 million after a video call in which every other participant was a deepfake.
Gartner GenAI attacks survey, 2025-09-22; Business Insider on Arup's loss, 2024-05.
Nothing here is a flaw in Marcus or in his assistant. It is structure. Authority delegates, and delegation concentrates. The assistant's inbox is efficient precisely because it holds everything, and that is exactly what makes it worth reaching. A patient adversary does not pick the lock on the front door. It reads the mail that says when the house is empty.
“I spent years hardening my own phone. The way in was my assistant's inbox, and I had never once thought about it.” Marcus, 46 — Master member
So the work followed the structure. With the company's written authorization, the assistant's systems came into scope. Her sign-in was hardened. Her recovery flow, the quiet back door into any account, was tested and closed. Approvals were given a second pair of eyes, so that no single message could move money. And, with her own consent and her own stop word, she practiced. The debrief taught; it did not shame. She now spots the shape of a pretext faster than Marcus does.
Marcus holds Master, $30,000 a month: monthly exercises across every channel, quarterly on-site testing, and 24/7 response.
A standing opposition, for one family.
A principal, her chief of staff, and a dedicated cell of five who never quite stop.
Eleanor runs a family office. There is real wealth, several households, a foundation, and a staff who move money and make arrangements every day. Her chief of staff, Daniel, knows where everything is. Between them they hold a surface too large for any single monthly check to cover.
The Second is not a tier you buy from a page. It is offered by introduction. It places a dedicated cell of about five people, plus the agent fleet, in standing service to one family. They behave as a patient, well-funded adversary would, and they do it entirely inside a signed consent envelope. Everything they are permitted to do is written down in advance, per person: which channels, which hours, what is always off limits, and the right to stop it all at a word.
The ground
In one survey, 43% of family offices had suffered a cyberattack in the prior two years. Close to a third had no incident response plan at all. The attacks are not rare. The plans are.
Deloitte Family Office Cybersecurity Report, 2024; a 2026 survey of offices managing about $119 billion found the same 43%. Ocorian, 2026-06.
What the cell actually provides is rhythm. Eleanor sees it as a calendar more than a service. The fleet runs every night, on local models, with deeper runs on frontier models for the hardest surfaces. Every month there is a bout: a human exercise, designed and supervised, that tests whether the family's protocols hold. Every quarter the cell comes on-site. They walk the houses and the office. They sit with new staff. They rehearse the bad morning before it arrives — your chief of staff's email is compromised; walk us through the next hour.
Because the same five people do this all year, they accumulate something no one-off engagement can. They know the family. They know that Daniel approves wires, and that Eleanor's daughter is about to take a public board seat, which changes her exposure. They know last quarter's findings, and they carry them forward, so each run targets what changed instead of starting over. The relationship is the product: menace, held entirely in service of the people it is pointed away from.
“I wanted one group of people who never stop imagining how we could be reached. And I wanted them to work for us.” Eleanor, 64 — The Second
Daniel describes it as the only part of his week that assumes the worst on his behalf, so that he does not have to carry that assumption alone. Eleanor describes it more plainly. She says it is like having an opponent who is on your side, which should not be possible, and is.
Eleanor holds The Second, by introduction: a dedicated cell of three to five, continuous, for one family. The number is named only on the membership page.
Five people, one idea
Every one of them is invented. Every one is someone we are built to serve.
The thread between them is small and old. A habit, practiced before it was needed, held by people on your side. That is the whole of it, from a grandmother's kitchen to a family office.