A chatbot waits to be asked. This one works all day.
The big chat apps only answer when you open them, and they forget you the moment you close the tab. A personal AI assistant works the other way around. It knows you, it remembers what you told it months ago, and it starts things on its own: a morning brief, a reminder before you need it, a heads-up before something goes wrong. You do not open it and ask. It works your day alongside you, built by hand for one owner.
We build the assistant around four parts of a life, because that is where the day actually happens.
It runs your work day
Your calendar, prep notes before a meeting, reminders that survive a chaotic day, and a running tally of what you promised whom, so nothing quietly slips through. It drafts the emails and notes; you approve what actually sends.
It keeps the people who matter close
Birthdays and anniversaries with enough lead time to do something about them, school events, gift ideas that don't arrive the day after. The nudge to leave by 5:15 so you make the recital, because it knew the recital was tonight.
It keeps you honest about your health
Not medical advice: general wellness encouragement only, and deliberately non-clinical. Workout and meal reminders, appointment tracking, streaks, and gentle accountability. It tracks and nudges; it never diagnoses or treats.
It catches you up every morning
A morning brief tuned to what you actually care about (your industry, your teams, your markets, your town), read to you like a sharp aide would, not dumped in a feed to scroll. You wake up already caught up, not with more to catch up on.
You don't configure a soul. You forge one.
Anyone can wire an API to a calendar. The personality is the part that takes hands. So the setup isn't a settings screen: it's a short, deliberate build, and by the end the assistant is yours, not a template with your name on it.
About an hour on your couch
The people in your life, the dates that matter, how your work runs, your health goals, the news you actually want, your humor, your pet peeves, and the one thing it should never do. Bring a partner or family: they'll remember what you won't think to mention.
We write the soul, you name it
Over the next couple of days we write the assistant's actual character (how it talks, when it pushes back, when it stays quiet) and seed its memory with everything from the interview. You pick its name. That part tends to sell itself.
By the end of the week it sounds like yours
The first week is tuning: too chatty, too formal, more sports and less politics, brief me at six not seven. By the end the personality is set. Then it's live, and the memory only compounds, with monthly tune-ups included.
Why ours don't lie to you.
We've run our own company on assistants exactly like this for months (three of them, with names and personalities), handling real operations every day. Everything that ever broke taught us a rule, and every rule ships inside the units we forge. This is the part a weekend tutorial can't copy, because it was paid for in real mistakes.
It verifies before it claims
No unit says a thing is done, booked, or sent until it has checked the actual result. An assistant that confidently lies to you once is finished, so ours are built to check first.
Its memory grows with you
Every correction sticks and every preference is written down in the same conversation you gave it. Month six knows things month one could not, and you can check that, because the memory is real files it writes and re-reads, not a feeling.
It drafts; you send
By default it prepares the email, the text, the reply, and stops there. It reads narrowly and drafts freely, because a bad draft costs you nothing, and anything going out under your name waits for you to say go.
It watches itself
It checks its own pulse on a timer, kills and restarts itself if it hangs, and has a second path to fall back on if that fails. You should never find out days later that your assistant has been quietly down.
Proof you can dial: (605) 315-3900; the AI answering our own phone runs on the same discipline your assistant would.
Your memories are yours.
Personal means intimate, so the privacy has to be real, not a paragraph on a policy page.
Isolated per client
Every assistant runs as its own instance. Your memory is never pooled with anyone else's.
Readable and exportable
The memory is a set of plain files you can inspect and export anytime. Leave, and you leave with them.
No training on your data
Your life doesn't become fuel for someone else's model. Your data trains nothing.
No advice posture
It tracks and reminds. It holds no medical, legal, or financial advice posture: it doesn't diagnose or advise.
Forged by hand, so the first group is small.
The service is new, so we are capping the founding cohort at ten, until the process is boringly repeatable and every assistant in it is genuinely good. Month to month, cancel anytime. The Soul Forge setup is waived for founding clients.
The presence in your pocket that texts first.
Founding-client rate: the interview and the forging, on us.
- Your own named assistant, forged for you
- Chat, plus a memory that keeps everything you tell it
- A morning brief, tuned to you
- Reminders that survive a chaotic day
- A news digest on what you actually care about
The full chief-of-staff.
Founding-client rate: the interview and the forging, on us.
- Everything in Companion
- Calendar and email integration (your accounts)
- Meeting prep briefs before you walk in
- The family module: dates, events, lead time
- The health module: nudges, tracking, accountability
- It starts things on its own, all day, without being asked
White glove. It talks out loud, and it runs on hardware set aside for you alone. The + covers voice usage, custom integrations, and that hardware.
Founding-client rate: the interview and the forging, on us.
- Everything in Executive
- A voice line: talk to it, not just text
- Custom automations and integrations
- Its own private setup, shared with nobody
- Priority tuning and support
Honest expectations
This isn't for everyone, and it's expensive to pretend otherwise. If you want a research tool you open when you have a question, a general chatbot already does that well for a fraction of the price. If you won't let it into your calendar and your day, most of the value never shows up: the magic is in the proactive part, and proactive needs access. And if you want something that acts entirely on its own, with no human in the loop, that's the opposite of how we build: ours prepare and wait for you, on purpose. If any of that is a dealbreaker, we'll tell you on the first call and point you somewhere better.
Tell us about your days.
An hour on your couch is the whole start. We'll listen, then say honestly whether we can forge you an assistant worth keeping, and if we can, you'll name it before you leave the room.