Guarantees
Four promises you can check: work resumes instead of repeating, an irreversible step can wait for you, an ungranted action is refused, and every run can be read.
Most of what an agent does is judgement, and judgement is not a promise anybody can make. These four things are different: Kortecx keeps them, they hold whatever the model decides, and you can check each one yourself.
Work resumes; it does not repeat
If a run of an app you started fails partway, you can retry it while the service that ran it is still up, and the retry picks up where it stopped instead of starting over: the steps that finished are reused, not run again. A run cut off when the service itself restarts is marked interrupted and is not resumed.
The part that matters is what happens to work that already reached the outside world. A message that was sent, a ticket that was filed, a row that was written: a retry does not do those a second time. A retry that would repeat one is refused, and a restart repeats it only after you confirm how many such steps it will run again.
Why this is the hard one
Retrying is easy. Retrying without re-sending the email is the difficult part. It is the promise on this page that matters most when an agent acts for you.
Anything irreversible can wait for you
An agent can read, think, search and draft on its own. On a run you gate, before it calls a tool that changes the world it stops and asks — and the run holds there until a person answers, or until its time runs out.
You choose which runs wait, per run or per trigger. See approvals.
An action outside what you granted is refused before it happens
Not undone afterwards — refused, before anything runs.
A model can propose any tool call it likes. Whether it happens is decided by Kortecx's own check against what you granted that run: the exact tool and the exact version, inside what your workspace's roles allow. Every one has to pass. The model's confidence is not an input to that decision.
A refusal is not a crash. The step settles as refused, with the reason, and you can read it back.
Every run is readable afterwards
You can see what an agent did, in order, after the fact: which step ran, what it reached for, what came back, what was refused, and where it stopped. Not a log you have to grep — a record the console renders.
See seeing what happened.
What these do not cover
The model's judgement is not one of the four
None of the above makes an agent correct. A run can resume perfectly, wait for approval politely, respect every grant, keep a complete record — and still have drafted a bad answer. The four guarantees are about what Kortecx does with an agent's decisions, never about the decisions.
They also say nothing about capacity, latency or availability. The full list of limits is in Limits.