A catalog of governed cases for real operations work. Rules, AI agents, and people work as a single system across the catalog. You decide what runs automatically and where a person steps in, with proof from end to end.
Browse the catalog below. Read any case in the docs.
Start from a vetted case, fit it to your operating policy, and run it without giving up control of the decisions that matter.
Start from a case built for the operational outcome you need.
Set its content, connect your systems, and name the people involved.
Rules handle what is known. Bounded agents reason where judgment is useful.
Your organization chooses what continues automatically and where a person steps in.
The run keeps its actions, decisions, hand-offs, and evidence together.
A catalog case becomes yours through its content, its connections, and its people. The catalog starts with a recommended human decision point. Your organization sets the final policy for its use case.
Thresholds, messages, templates. The numbers and words the case runs with.
The outside systems the case reads or writes, or the Hyphen default when there is nothing to connect.
Who reviews or approves when your policy calls for a person, and who receives a hand-off when an agent reaches its boundary.
Every door uses a governed Hyphen run path and produces a record.
A person opens a configured case and runs it. When its policy calls for a decision, the task waits for the right person.
Start the configured case from your backend with one API call. Any configured decisions still pause for the people you named.
An assistant starts a permitted catalog playbook through the MCP server. It can choose the playbook, but it cannot change its steps or declare new tools.
Each case says what it closes, its recommended human decision point, and whether it can run on Hyphen defaults or needs setup first.
No case matches those filters.
Write it the way you would tell a new colleague: what comes in, what should happen to it, and who decides. Hyphen drafts the case, checks it, runs it on sample data, and lists it as yours.
Book a 30-minute demo and we will run a case against a sample of your data.