The same workers, without the screen.
Every Hendl worker can be called headlessly — from your portal, your queue, your scripts or your own agents, over MCP, A2A or plain REST. Same playbooks, same gates, same audit trail. No interface required.
Infrastructure, not just an app.
Your case operations shouldn’t be trapped behind a login. Hendl’s workers are callable capability — from anywhere you build.
Workers run without the UI — called by your systems, your scripts or other agents, doing the same governed work.
MCP and A2A native. Hendl’s workers can be tools in your agent stack, and your agents can be workers in Hendl.
Headless doesn’t mean lawless — the same permissions, gates and audit trail apply whoever makes the call.
Start a coverage check from anywhere.
A single request runs the same coverage-check worker your team uses in the app — citations, autonomy limits and audit reference included in the result.
Speaks the protocols your agents do.
Hendl is a first-class citizen of the agentic web — a tool server to your agents, a peer to your partners’, and plain HTTP to everything else.
Every Hendl capability — case search, document intake, coverage checks — exposed as MCP tools your agents can call.
Hendl workers advertise themselves as A2A agents: discoverable, negotiable, callable by any compliant peer.
Plain HTTP for everything — start work, stream progress, receive results. No SDK required.
Point a queue at a worker and process a portfolio overnight — same playbooks, no hands.
Where headless earns its keep.
Your customer portal calls a worker to triage a claim at the point of submission.
Your in-house agents call Hendl workers as MCP tools for the regulated steps.
A queue feeds ten thousand historic cases through a review worker overnight.
A counterparty’s A2A agent negotiates document exchange with yours — logged on both sides.
Questions builders ask.
What exactly is a headless worker?
The same digital worker that runs in the Hendl UI, invoked programmatically — by API, MCP or A2A — with no interface in the loop.
Is headless work governed the same way?
Identically. Permissions, autonomy levels, approval gates and the audit trail are enforced at the worker, not the UI — there is no back door.
What is MCP support, concretely?
Hendl runs an MCP server: your agents connect and see Hendl capabilities as tools — case search, intake, checks, drafting — with scoped credentials.
And A2A?
Hendl workers publish A2A agent cards, so external agents can discover them, negotiate a task and receive results — with the full exchange logged.
Who approves gated actions on a headless run?
The same named humans. A gated step pauses the run and routes the approval to Hendl — the caller polls or receives a webhook when it resolves.
How do we authenticate callers?
Scoped service credentials per integration, rotatable and revocable, with per-worker allowlists. Every call is attributable to its caller.