Casework runs on judgement. Everything else is overhead.
Every claim, dispute and investigation is decided by a person weighing evidence. Yet most of a caseworker's day is spent feeding systems, chasing documents and re-keying facts. We started Hendl because that ratio is a choice — and the industry keeps choosing wrong.
Six beliefs, built into the product.
The case is the unit of work.
Not the ticket, not the task, not the inbox. Casework has parties, evidence, deadlines and consequences — software that ignores that shape forces people to carry it in their heads.
Autonomy is a dial, not a switch.
Between “a human does everything” and “the machine does everything” sit a hundred useful settings. Teams should choose one per task — and change their mind with evidence.
Every action must answer for itself.
Who did what, when, on whose authority, based on which evidence. If an action can’t answer, it shouldn’t happen. That is what regulated work means.
Humans decide what matters.
Agents draft, gather, reconcile and chase. People weigh, judge and own outcomes. The org chart doesn’t change — the ratio of judgement to drudgery does.
Software should do the work, not track it.
A generation of tools asked caseworkers to feed the system. We think the system should feed the caseworker — work arriving already advanced, with the evidence attached.
Trust is earned in the audit.
Anyone can demo magic. The test is eighteen months later, in front of an ombudsman, when every decision replays cleanly. We build for that day.
What we won't build.
A manifesto is only honest if it costs something. These are the lines we hold, even when they lose us a deal.
If we can’t show the reasoning, we don’t ship the feature.
Charging per seat for software whose whole point is doing more with the same people.
Your cases export completely, in open formats, on any day you ask.
A text box is not a digital worker. Agents here finish work, not sentences.
Provenance is captured at write time or it isn’t real. It can’t be reconstructed later.