Built by practitioners. For the work IT really does.

Who this is for

Three people, one record.

We built this for teams that need support, maintenance, and project work to move without losing who authorized it, what changed, or how the target verified it.

Why the company exists

AI got good at doing the work before anyone solved checking it.

The industry answered "can we trust the AI?" with better prompts, longer transcripts and a confidence score. Those are all the model talking about itself. The only answer that survives a hostile question is an independent one: the machine that was changed says whether it changed.

• Standing privilege was the original sin. Automation needed rights, so it got permanent rights, and every audit since has been an exercise in explaining an account nobody can remove.
• The evidence and the actor were the same party. A tool wrote a line about its own work, and that line became the audit trail. Nobody designed that; it accumulated.
• AI made both problems worse at once. Now the thing acting is non-deterministic, holds standing rights, and writes a fluent, confident account of what it believes it did.
• So we inverted it. Privilege is leased and expires. The model never receives a credential. The target machine answers. The undo is written first. What comes out is a record, not a narrative.

This is a young company building an ambitious product, and we are not going to pretend otherwise on the page where we introduce ourselves. What we can defend is the architecture and everything that is already running — and we would rather you tested both than took either on trust. The trust center →

Constraints we chose

What we will not build.

These are not warnings about a future release. They are decisions, and each one costs us something in a sales call.

• A mode where nobody's name is on the work. Autonomy changes when a human authorizes, never whether a human is accountable. Someone is always on the record.
• Metered proof. Review the published package scope and evidence capabilities with your operating requirements before purchase.
• A certification we have not finished earning. SOC 2 Type II is on the roadmap and the report is not in hand yet; until it is, we hand you the evidence directly instead of a badge. Where we stand on assurance.
• A claim we cannot demonstrate. If it cannot be shown on a live system, it comes off the site — including off this page.
• Invented proof. No customer names we do not have, no fabricated case studies, no metric we cannot source. The illustration on this site is a printer at a medical practice, and it is clearly an illustration.
• Unclear enrollment allowances. $200 per tenant plus $299 per named technician monthly, with 30 enrolled endpoints per technician. Review the allowance before purchase.