Introducing Omni - ARGOS Identity

ARGOS Omni is a Review Ops Agent that understands your company’s own business process and runs it. Every company has its own process and its own criteria for deciding. Omni understands them and performs the review work people used to do — checking documents, comparing information, and applying criteria to reach a decision. Your Process. Your Rules. Your Agent.

The process lives in people

At many companies, the processes that matter most still depend on “the person who knows how it works.”

“How do we handle this case?”
“Why did we reject it last time?”
“Do we need to check this document too?”
“Is there a record of what this decision was based on?”

Even with an SOP, real work produces countless exceptions and judgment calls, and the harder a case is, the more surely it lands on the most experienced reviewer. The Process Lives in People.

Why review is hard to automate

Review work is not simple repetition. It means understanding a company’s Process · Rules · Know-how and deciding with them.

What needs to be checked
What needs to be compared with what
Which criteria apply
When a case should be handled as an exception
And why the decision was made

Review is not just data processing. It is the execution of the business knowledge a company has built up.

How a Review Ops Agent works

Omni understands that Process · Rules · Know-how and works exactly the way your company does.

Understand

Understands your business process and decision criteria. It reads your plain-language policy and output schema and builds a playbook of verification steps, the engines they need, and their execution order.

Execute

Reviews, verifies, compares, and decides the way your company does. It runs the playbook actions in order — checking documents, extracting data, cross-checking documents against each other, and running external lookups.

Explain WHY

Records the Rule · Evidence · Reason behind every decision. Each extracted value carries the reason for it (reasoning) and the step it came from (sourceStep), and the playbook as it stood at run time is kept with the result.

Understand your Process → Execute your Process → Explain WHY Here is one policy clause going through those three stages.

Policy:  "The account holder on the bank verification letter must match
          the entity name on the certificate of good standing."
   ↓  Understand
Action:  verify_bank_holder  (engine: Text Verifier)
   ↓  Execute
Result:  bank_account.matches_company = true
   ↓  Explain WHY
Reason:  "The account holder 'ACME TRADING LLC' on the bank verification
          letter matches the legal name on the certificate."

From operator to supervisor

Bringing in a Review Ops Agent turns people-centered review into agent-centered operations.

	Before — Human Operator	After — Review Ops Agent
Who does the work	People review and decide every case themselves	Omni performs the review according to your process
Consistency	Varies with each reviewer’s experience and skill	The same rules and criteria are applied every time
Grounds for decisions	Missing, or recorded differently by each person	The WHY and evidence of every decision are recorded
Audits and follow-up	Past records are dug up to reconstruct the decision	Past decisions and their grounds can be traced immediately
Work output	People write the result and enter it into systems	AI Agent Report + Customized API
Role of people	Operator — does the work	Supervisor — oversees and manages exceptions

The organization shifts from one where people run the process to one where Omni runs the process and people supervise it. People handle the cases decided as needing review (pending_review) and the exceptions, and oversee the grounds for each decision.

Two outputs

AI Agent Report

Decision · Findings · Evidence · WHYFor supervision, audit response, and follow-up. View it on the analysis results screen in the dashboard, and get it as a PDF from GET /analyses/:analysisId/report.

Customized API

Decision · Data · Reason · Evidence · Next ActionConnects straight into the next task in your existing systems. GET /analyses/:analysisId returns the decision (verificationStatus) and the data assembled to your output schema, with reasoning and sourceStep on every field. When you need a next action, define it as a field in the output schema and it comes back with the rest.

From traditional automation to agentic ops

Traditional automation broke work into pieces and wired them together. Omni understands the whole task and performs it.

	OCR	RPA	Omni — Review Ops Agent
How the work is done	Extracts the information needed	Runs predefined steps and flows	Understands the context of the whole task and performs it
How decisions are made	People judge the extracted results	Branches on predefined rules	Decides from your process, rules, and context
Exceptions	Handled by people	Needs separate rules and exception flows	Understands each exception as a case and handles it
Adapting to change	Fields and templates are modified	Rules and flows are modified	Reflects changes to your process and rules
Operations	Ongoing upkeep per template	Ongoing upkeep per bot and flow	Run as one integrated agent
Final output	Data	Action	Decision + WHY + Evidence

Traditional Automation

OCR + RPA + Rules + Exception Flow + HumanSplit → Define → Connect → Maintain the work.

Agentic Ops

Omni — Review Ops AgentUnderstand → Review → Decide → Explain the work.

From Fragmented Automation to Agentic Ops.

Use cases

Every company’s process is different, but any work where people review and decide is work a Review Ops Agent can take on. The examples below show how. The review steps become playbook actions following the procedure in your workflow policy, and the decision comes back as a verificationStatus value.

Telecom
Financial · KYB
Insurance

Corporate telecom subscription review

Item	Details
Input	Business registration certificate · Certificate of incorporation · Authorized signatory’s ID · Power of attorney · Service application
Review steps	Check the documents → Extract the data → Cross-check across documents → Verify signing authority and the power of attorney → Apply the subscription review rules → Decide on exceptions
Decision	Approve (`verified`) · Reject (`rejected`) · Needs review (`pending_review`)

Business customer onboarding review

Item	Details
Input	Corporate documents · Shareholder information · Representatives · UBOs · External data
Review steps	Confirm the entity information → Verify relationships → Cross-check against external data → Check risk → Apply the onboarding policy → Decide on exceptions
Decision	Approve (`verified`) · Reject (`rejected`) · Needs review (`pending_review`)

Insurance claim review

Item	Details
Input	Claim form · Medical or accident records · Policy information · Customer information
Review steps	Confirm the claim information → Verify the required documents → Cross-check against the policy → Apply the payout criteria → Decide on anomalies, omissions, and exceptions
Decision	Pass (`verified`) · Reject (`rejected`) · Needs review (`pending_review`)

Different Process. Different Rules. Same Review Ops.

By verification type

These are the verification types most often used in a workflow.

Type	What it does
KYB	Confirms the legal entity and its beneficial owners, and verifies that the submitted documents agree with each other
AML	Checks individuals and companies against sanctions and watchlists
Document compliance	Checks that every document a regulation requires was submitted, completely and correctly
Address verification	Confirms that an address on a document exists, and normalizes it
Income verification	Extracts figures from financial documents and checks consistency across them

Coming soon: the KYC (individual identity verification) capability of ID Check will be integrated into Omni. If you need KYC today, use ID Check.

Next steps

Quickstart

See the whole flow in How to Use, then build a workflow and run your first analysis end to end.

Core concepts

The data hierarchy from project to analysis, plus playbooks, verdicts, and scores.

Creating a workflow

A real walkthrough, including the policy text.

Authentication

Getting an API key and using the x-api-key header.

​The process lives in people

​Why review is hard to automate

​How a Review Ops Agent works

Understand

Execute

Explain WHY

​From operator to supervisor

​Two outputs

AI Agent Report

Customized API

​From traditional automation to agentic ops

Traditional Automation

Agentic Ops

​Use cases

​By verification type

​Next steps

Quickstart

Core concepts

Creating a workflow

Authentication

The process lives in people

Why review is hard to automate

How a Review Ops Agent works

From operator to supervisor

Two outputs

From traditional automation to agentic ops

Use cases

By verification type

Next steps