CONFIDENTIALThe agentic operating system for your professional and personal life.
Pax does the work, finds the work you can hand off, and gives you back the hours for the work only you can do. One assistant, one memory, every channel. Pax does the work. You approve.
An operating system can run the day itself.
The thesis below is the company's, as written. This overview walks it as an argument in three parts. The day can be run by a system that watches how your work actually moves. That system hands you only the moments that need your judgment; everything else it does, and you approve. And it gets measurably better every time you touch it, which is the personalization thesis the beta will test.
The next AI breakthrough is not another model. It is intelligence reliably accountable for getting work done. Paciva exists to manage the complexity between intent and outcome: what comes in, what matters, what is known or assumed, what may be done, and where intelligence should run. Work moves forward under your rules and returns only as a result, blocker, or decision, giving you time back without taking control away.
Paciva is the user-controlled execution layer between abundant intelligence and trusted outcomes. That is the destination. A new beta cohort is underway; the first saw limited usage of an earlier product variant. Product examples show the intended experience, not customer results.
It runs the day
An operating system that watches how your work actually moves and proposes the handoff from observed reality.
You keep the judgment
Pax does the work. You approve. Only the moments that need your call reach you, with the reason attached.
Every touch compounds
Every action is labeled behavior. The system gets measurably better each time you touch it, tuned to you alone.
Your day is spent coordinating work instead of doing it. Nobody is watching the leak.
Microsoft reports 275 daily pings for the top 20% of users by ping volume, with roughly two minutes between pings during work hours. That telemetry is distinct from its 31,000-person survey. Teams spend 50% more time in unnecessary meetings than on their highest priorities. Two leaks run at once: graymail has a modeled time value of $12,796 per employee per year, and the Collaboration Tax, the operational drag of meetings, tool switching and AI rework, has a separate modeled value of about $12,800. The leak persists for one reason: nothing in your stack watches how your work actually moves. That is the job an operating system exists to do.
One assistant across eight modes, entering another beta cohort.
The product direction spans eight modes, all the same Pax: the same memory, trust, and rules follow you from the front door to the calendar to multi step work running across your tools. Every mode obeys the same contract from the thesis: Pax does the work, and the moments that need your judgment come to you as one decision, not eleven threads.
The assistant is the surface. The operating system underneath finds the work you can delegate before you ask.
This is the first beat of the thesis made mechanical: an operating system can run the day because it watches how the day actually runs. Its recall runs beyond any human on the team, one pane across every connected tool, and a memory that does not walk out the door when a teammate does.
On the other hand: the Defense Center
The operating system finds work and hands it off. The Defense Center blocks what should never have reached you and shows its work: every action with its reason, reversible ones undone in one tap, irreversible ones always gated. That ledger is what earns Pax the right to act for you.
The dial: Pacivity
One dial sets how boldly Pax acts: Suggest, Act and you approve, Standing autonomy. The confirmation gate is architecture, not policy. No waiver can move an irreversible action past it.
Personalization is a technical thesis. Adoption is the next test.
Every action inside Paciva is labeled behavior: who matters to you, what gets approved, which threads are dead, how you decide. Two mechanisms carry the argument, and the third beat of the thesis is also the clock: a system that gets better with every touch makes early seats sticky before the category consolidates, and makes every month of delay a month of calibration a rival never gets back.
The reported technical result is 8.5% false positives on day 0 and 0.41% by day 90. Cohort size, dates, evaluation set and reproducible methodology are needed to validate this comparison. It does not establish customer retention, willingness to pay or paid adoption.
A crowded field. Our bet is governed work across channels.
Email assistants, general-purpose AI and automation platforms already cover parts of this job. Our differentiation thesis combines cross-channel context, screening, and execution under explicit user control. The beta must show why early-adopter knowledge workers and developers choose that combination over their existing tools.
Priced like software. Valued like labor. Expands with the work.
Our first buyers are early-adopter knowledge workers and developers. Pricing is proposed and untested: $300/month founding access, grandfathered, and $3,000/month list, plus metered usage. The next beta tests recurring value and conversion before we claim a repeatable acquisition model.
We are raising $2M.
Use of funds: acquisition, compliance, compute, and hiring. The next system and beta-to-paid transition are the focus. Terms and planning detail are in the full brief, alongside the market model.
Built by two people. A new beta is underway.
Our own engineering runs on autonomous agents with governed review, measured by accepted merges and review gate pass rate rather than raw PR count. The first external cohort saw limited usage of an earlier variant. The new cohort is the next test of engagement and paid demand; internal engineering usage is separate evidence.
Jeremy Mays
Two decades across B2B and B2C growth on brands including American Express, Comcast, Four Seasons, Koch, Sotheby’s, Guaranteed Rate, Affectiva and the Los Angeles Police Department. $120M in new ARR driven in 2.5 years, scaling a team from 1 to 24.
Frederick Townes
Created W3 Total Cache, running on tens of millions of sites. Founded W3 EDGE, serving Coca-Cola, Microsoft, Sony, and Staples. Led product at Remine across roughly 60 MLS markets and 1.2M professionals. COO through acquisition at Hyperlift.
The evidence is in the full brief.
Example product walkthroughs, reported technical results and their evidence gaps, market assumptions, the beta plan, and the ask. Pax reads along and answers what is still open. It takes a name and an email at the door.
Rather see it than read about it?
Thirty minutes with the founders, Pax running live on real inbound. No slides.