🤫husshhussh
🤫husshhusshOnePuppy
🤫 Talent 2.0 · August 2026

How we hire, and what we owe you.

This is the whole handbook. It is short on purpose: what matters to somebody applying is what we promise and whether we keep it, not how our internal process is organised. Every promise here applies to everyone who applies, from the first minute.

See open rolesWhy we are building this

careers@hushh.ai reaches a person who works here.

  1. 01We begin with the human
  2. 02What we promise you
  3. 03How hiring works here
  4. 04Michelle
  5. 05How we use AI in hiring
  6. 06Your information
  7. 07If you need something different
  8. 08Ways to work with us
  9. 09Reaching a person

We begin with the human

Most technology companies begin with the company. What can we collect, what can we predict, what can we sell. We begin at the other end, with the person, and we ask what they would want if they understood exactly what was happening.

That is not a value we put on a wall. It is the product. 🤫 Private Agent One is an AI that answers to one person, runs on hardware they own, and shares what they allow and nothing else. It is free to every American for life, which means the money can never come from them.

We hire people who find that worth doing, and who would find the opposite hard to do.

What we promise you

Six things. They are unconditional, they apply to everyone who applies, and none of them depends on how far you get.

  • A person reads your application. Not a filter, not a score.
  • You get an answer either way, and a reason if you want one.
  • You always know your stage, who owns it, and what happens next.
  • Any work sample is paid, time-boxed, and never used in our product.
  • No AI ranks you, hires you, or declines you.
  • Ask us to delete everything we hold about you and we delete it, without asking why.

How hiring works here

Short, and the same for everyone.

You apply in a few minutes. No cover letter, and no form that takes longer to fill in than the job takes to explain.

If we are going forward, we ask you to sign a confidentiality agreement before we show you anything that is not public. You can have a lawyer read it, and you keep a copy. If you would rather not sign, say so and we will stop there without hard feelings.

Then conversations: the person who owns the role, someone who will judge the craft, a piece of representative work you are paid for, a working session with people you would actually work with, and a conversation with Manish. You see the questions and the rubric before each one.

Then an offer or a no, in writing, from a person, with a reason.

We aim to move quickly and we will tell you what we are waiting on. Senior, regulated, or immigration-dependent roles take longer, and you will hear why rather than hearing nothing.

Michelle

Michelle is our recruiting agent. She is software, not a person, and we say so the first time she writes to you. She keeps track of where every candidate stands and writes to you when something changes, on the day it changes.

Every message tells you three things: what just happened, the one thing you need to do, and when you will hear next. If email fails, the same message is waiting on your own page, so a bounce never costs you the knowledge of where you stand.

Reply to her and a person reads it.

How we use AI in hiring

We build AI. We are careful about pointing it at people.

AI helps us schedule, draft, summarise our own notes, and keep candidates informed. That is Michelle, and it is why nobody here waits in silence.

AI does not rank candidates, score them, decide who advances, or decide who is declined. Those are made by named people who write down what they saw. A machine is not going to be the reason you did not get a job here.

We do not use candidate material to train a general-purpose model. We do not record an interview without asking first, and there is always a version of the conversation that is not recorded.

You are welcome to use AI on a work sample. Tell us how you used it. Using good tools well is the job; pretending you did not is the only thing that would count against you.

Your information

We keep what you send us and the record of how the process went, and nothing else. We do not buy data about you, we do not build a profile, and we do not share your application with anyone outside the people deciding.

You can correct anything factual, withdraw at any time, and ask us to delete everything. Deletion is immediate and needs no reason. Where the law requires us to keep a record, we will tell you which record and why.

None of this is a condition of applying, and using any hushh product is never a condition of applying either.

If you need something different

Tell us what would work and we will do it. A different format, more time, a written conversation instead of a spoken one, a different time of day.

You do not have to explain why, and asking is not part of the assessment.

Ways to work with us

Not every good arrangement is a job. We work with full-time employees, interns, contractors, consultants, ambassadors, and creators, and each has its own terms.

In many places an engagement starts with a paid probation or a paid first piece of work before it is confirmed. Where that applies to you, we put the length, the terms, and what confirmation depends on in writing before you start. Never after.

Reaching a person

careers@hushh.ai reaches a human being who works here. Not a queue, not a form, not a bot.

That includes if you think we got something wrong, if you want feedback on a decline, or if you want to be removed from everything.

About this version

Talent 2.0 replaces Talent 1.0, which is withdrawn and no longer available for download. 1.0 published our internal operating manual, including drafting on confidentiality agreements, background checks and worker classification that we are not qualified to write. Keeping it up with a note would not have fixed that. Every promise 1.0 made to a candidate is on this page, and the operating detail is now where operating detail belongs.

See open rolesAsk us anything