Product overview

One work platform for every insurance employee.

ASD-STE100 Simplified Technical English No external libraries Responsive

WorkForceOS is a work platform for insurance employees. It is not a new portal. It is not a chatbot. It is not one more app.

WorkForceOS connects the employee, the data, the business systems, and the automation in one work environment. The employee sees the work. The AI helps. The system does the approved steps. The control layer checks every important operation.

1. The four systems

Use one name for each layer. Do not mix the names.

System	Serves	Primary user	Main goal
InsureOS	The customer	Personal agents, customers	Runs insurance transactions. Quote, buy, renew, claim.
WorkForceOS	The employee	Employees, teams, managers	Runs internal work. Tasks, processes, automation, control.
EmployeeOS	The employee	One employee	The user layer of WorkForceOS. Tasks, cases, assistant, approvals.
ControlOS	The organization	Risk, compliance, audit	The governance layer. Rights, rules, approvals, audit log.

Triglav appears in this proposal only as an example of a large insurer. Confirm all integrations and priorities with an analysis of the real environment.

2. The structure of WorkForceOS

The platform has a user layer and a service layer. Click a module. Read what it does.

EmployeeOS — the user layer

MyWork — the personal start page

The employee sees the tasks, the deadlines, and the open cases. The system proposes priorities from deadlines and business rules. The employee can change the priorities.

Work Assistant — the AI helper

The employee asks in plain words: "Summarize this case" or "Prepare a reply to this message". The assistant uses approved internal sources. It shows the source of each fact. It says when it does not know. It never invents rules, decisions, or customer data.

My Cases — the case view

Each case has one ID, one owner, a state, a deadline, the documents, the history, and the next step. The employee sees the whole flow. The employee does not search many apps.

My Approvals — the approval view

The employee sees the work that waits for approval. The employee approves or rejects. The system records each decision.

Platform services

KnowledgeOS — knowledge and procedures

The employee searches internal instructions, policies, work procedures, and product documents. The answer comes with a summary and links to the sources. The system respects access rights.

Document AI — documents and data

The system reads documents. It finds the type. It extracts the data. It checks the completeness. It prepares summaries.

Workflow Engine — process automation

The engine connects single steps into a full process: receive, classify, extract, check, prepare, approve, record. It can run the allowed transactions when the rules permit.

TeamOS — handovers between teams

The system prepares a structured summary for the next team. It assigns the task to the right team. It warns about stale cases. It shows the workload.

ControlOS — governance

ControlOS manages the rights of employees and AI agents. It sets the limits for automatic decisions. It requires approvals. It keeps the audit log. It is part of the platform from day one, not an add-on.

3. The process flow

This diagram shows the first pilot process: a document moves from arrival to the business system. Every step has one owner and one result.

Employee / input AI step Decision Approved execution Missing data path Record

Document arrivesfrom an approved source

Classify documentDocument AI finds the type

Extract dataDocument AI reads the fields

Documents complete?

Create task:
request missing documents

YES

Prepare case summarywith links to the sources

Employee review

APPROVE

Execute allowed operation

REJECT

Return to work queue

Update business system

Audit logevery step, every decision, one chain

View the Mermaid source of this diagram

The diagram above renders this Mermaid definition. The page uses no external libraries, so the site renders the diagram itself. Paste the source into any Mermaid tool to get the same chart.

flowchart TD
    A([Document arrives]) --> B[Classify document]
    B --> C[Extract data]
    C --> D{Documents complete?}
    D -- No --> E[Create task: request missing documents]
    D -- Yes --> F[Prepare case summary]
    E --> F
    F --> G{Employee review}
    G -- Approve --> H[Execute allowed operation]
    G -- Reject --> I[Return to work queue]
    H --> J[Update business system]
    I --> J
    J --> K[(Audit log)]

    style A stroke:#3b82f6
    style B stroke:#a78bfa
    style C stroke:#a78bfa
    style D stroke:#f59e0b
    style E stroke:#f87171
    style F stroke:#a78bfa
    style G stroke:#f59e0b
    style H stroke:#34d399
    style I stroke:#3b82f6
    style J stroke:#34d399
    style K stroke:#94a3b8

4. Three modes of operation

An AI assistant is not the same as automation. Separate the modes. Apply strict control to the risky work. Select a mode:

5. The ControlOS policy

ControlOS sorts every operation into one class. The class sets the control level.

Automatic

The system acts. It records the action.

Summarize a document
Check completeness
Classify a case
Create a task
Prepare a draft

Needs approval

The system prepares. The employee approves.

Send a reply to a customer
Update a business record
Close a case

Forbidden for AI

A person decides. The system prepares the basis.

Decide about coverage
Reject a claim
Pay a claim

6. The first 90 days

Do not build the full platform first. Prove the concept. Then grow. Click a phase:

Weeks 1–2 — Talk to the employees

Interview employees and team leaders. Ask about repeated tasks, copied data, searches, and waiting. Result: a list of the main problems.

Weeks 3–4 — Map the processes

Analyze the processes, the systems, and the data. Result: a task catalog and a map of the system connections.

Weeks 5–6 — Rank the opportunities

Score each task: frequency, time, standardization, data access, risk. Result: the priority use cases.

Weeks 7–8 — Build the prototype

Build MyWork and the Work Assistant. Result: a clickable user interface.

Weeks 9–10 — Build one automated process

Automate one selected process end to end. Result: a working proof of concept.

Weeks 11–13 — Test and decide

Test with real employees. Measure time, quality, and usability. Result: a decision about the next step.

7. The productivity metrics

WorkForceOS is not a tool to watch every click. The goal is to remove unneeded work, not to count messages.

Metric	What it measures
Completion time	Time from start to end of a task
Manual work	Time spent on repeated steps
Automation rate	Share of suitable tasks with no manual steps
Quality	Share of results with no errors
Rework	Cases that come back because of missing data
Waiting time	Time a case waits for data or approval
Employee satisfaction	The effect of the platform on the work

Value calculator

Change the numbers. The calculator shows the saved time. Confirm the result with a pilot.

Tasks per employee per day

Minutes per task today

Minutes per task with WorkForceOS

Number of employees

Saved time does not mean fewer employees. It can mean more cases, better service, or more time for complex work.

8. The first pilot: Document & Case Copilot

This pilot tests the search, the AI, the integrations, the orchestration, and the control in one process.

The system does

Takes the documents from an approved source
Finds the document type
Extracts the important data
Searches the connected systems
Prepares the case summary
Checks the document completeness
Prepares the next task
Records the result

The employee does

Reviews the summary
Corrects the draft
Approves the next operation
Decides the hard cases

Start mode: the system only proposes. It does not change business records. Add allowed operations later, when the quality is proven.