Agentic · Agent Hub

Agent Hub is where the work happens.

Shelbe answers questions. Agent Hub takes actions.

Together they close the operational loop: conversation in, structured data extracted, routed to the right system of record, action committed, status returned to the worker.

Every consequential action is gated by human-in-the-loop dependencies and logged in Scout — the orchestration plane that turns agentic AI from advice into governed, completed work.

See the workflows Talk to leadership →

Human in the loop, live

AI proposes. You approve. Everything is logged.

Every consequential action runs the pipeline to a governance gate and waits for a named human. Approve or reject one below — and watch the Scout audit trail capture who proposed it, who decided, the policy it was checked against, and what happened.

Intent extraction
Structured data
System routing
Human in the loopyour call

The use-case library

One governed loop. Nearly unlimited use cases.

A sample of what runs on RedeApp today, across every frontline function. Open any one to see how the loop carries across roles — the same governed loop carries each one, and the ones you bring next.

Loading the use-case library…

HIPAA-compliant PHI messaging

Care teams are already sharing PHI. Make it compliant.

In senior living and healthcare, frontline staff are almost certainly sharing protected health information on non-compliant consumer apps today — a real legal exposure. RedeApp keeps that messaging inside a HIPAA-compliant, encrypted, governed channel, so caretakers share mission-critical resident information with exactly the right people.

Used alongside required clinical documentation, not as a replacement — state regulations vary.

“[Resident]’s daughter requested no more pull-ups — briefs with pads only.”

“[Resident] is C. diff negative.”

“New admit [Resident] is now on nectar-thick liquids and a mech-soft diet.”

“All staff: please turn [Room ###] every 2 hours, left and right, with pillow support — the wound worsened over the weekend.”

“[Resident] now requires a Hoyer for bed transfers; sit-to-stand for toileting until the toileting pad arrives.”

The four layers of Agent Hub

How orchestration actually runs.

Each layer below is independently configurable, audit-logged, and customer-controlled.

Intent extraction

Shelbe identifies the intent of the conversation: shift swap, maintenance request, procurement order, time-off request, safety report.

Each intent has a defined structured-data schema and a routing destination.
Structured data collection

Conversational data extraction: the worker says

'I need to swap my Friday shift';

Agent Hub identifies the date, the shift, the requested replacement, and the policy constraints (overtime, scheduling rules). Missing fields are clarified conversationally.
System routing
Once structured data is complete, Agent Hub routes to the right system of record via the integration library:
- Workday
- ADP
- ServiceNow
- SAP
- Azure
- etc...
Authorization Forwarding ensures workflow ownership stays with the source.
Maker-checker governance
Consequential actions enter human-in-the-loop approval workflows.
- Manager approval for shift swaps.
- Supervisor approval for procurement above threshold.
- Audit-logged decisions with policy references.
- Full Scout trail.

Live Today

Not demos. Real workflows, at real customers, running now.

Operational workflows are in production today for RedeApp customers:

shift swap and labor fulfillment
maintenance and asset management
procurement and inventory orchestration
onboarding and pre-hire velocity
knowledge retrieval at the line.

Each is documented with the system integration, the maker-checker pattern, and the customer reference.

See the workflows →

Questions

Frequently asked questions

How is Agent Hub different from typical advisory enterprise AI?

Most agentic AI in enterprise software is advisory — it tells a human what to do, and the human still does it. Agent Hub instead routes the agent's proposed action through the actual system of record, applies maker-checker governance, and returns a completed status to the worker.

What are the four layers of Agent Hub's architecture?

Agent Hub runs on four layers: intent extraction (identifying what the worker needs, like a shift swap), structured data collection (conversationally gathering missing details), system routing (sending the request to Workday, ADP, ServiceNow, or SAP), and maker-checker governance for approval.

What is maker-checker governance in Agent Hub?

Maker-checker governance means consequential actions — like a shift swap or a procurement request above a threshold — enter a human-in-the-loop approval workflow, with the approving manager's decision, and the policy it was checked against, fully logged in the Scout audit trail.

What live workflows does RedeApp say are running through Agent Hub today?

RedeApp names shift swap and labor fulfillment, maintenance and asset management, procurement and inventory orchestration, onboarding and pre-hire velocity, and knowledge retrieval at the line as operational workflows in production, each documented with its system integration and maker-checker pattern.

What's the relationship between Shelbe and Agent Hub?

Shelbe handles the conversation — understanding what a worker is asking for — while Agent Hub takes the resulting proposed action, routes it through the appropriate system of record, applies governance, and closes the loop by returning a completed status back to the worker.