Use cases
What people actually build, at the level of mechanism.
Each use case names the trigger, the systems involved, how the workflow works, where a human decides, and what the outcome is. No stand-ins for a real conversation about your process.
- use cases across three areas
- 12use cases across three areas
- human control point in every case
- Namedhuman control point in every case
- solution page for each
- Linkedsolution page for each
- for a scoping conversation
- No stand-insfor a scoping conversation
Browse by industry
Industry use cases built from real system combinations.
Each runs two or more connected systems in one governed process, with a named decision point and a complete record.
Industry use case: a business process in a specific industry, built from two or more systems.
Integration pattern: the architecture for one system combination. Browse patterns
Industry
Theme
10 use cases
Public sector
Cross-agency file intake and validation
Microsoft Azure + SAP + ServiceNow
Files from other agencies are validated before they touch finance records, and problems become tickets.
Defense & national security
Cloud incident to controlled remediation
Amazon Web Services + ServiceNow + Jira
Cloud events become incidents and engineering tasks, and remediation runs only when approved.
Defense & national security
Configuration change across controlled environments
ServiceNow + GitHub + ArgoCD
Configuration changes deploy only after approval, with checks and deployment events tied to the change record.
Critical infrastructure
Infrastructure alert to incident and remediation
Amazon Web Services + ServiceNow + Jira
Alerts become incidents and engineering issues, and infrastructure changes run only when approved.
Critical infrastructure
Deployment governance
GitHub + ArgoCD + ServiceNow
Every deployment links to an approved change, passing checks and a deployment event.
Critical infrastructure
Cloud event to analytics workflow
Amazon Web Services + Snowflake + ServiceNow
Cloud events are enriched, loaded for analysis and escalated when thresholds are crossed.
Enterprise
Alert classification and grouping
Amazon Web Services + ServiceNow + OpenAI
Operators receive proposed incident groups with their underlying AWS alerts preserved and the grouping rationale recorded.
Enterprise
Proposed remediation for an incident
ServiceNow + Amazon Web Services + OpenAI + Jira
Incident owners receive a source-backed remediation plan ready for technical review and the required change approval.
Enterprise
Change summary for approvers
GitHub + ServiceNow + OpenAI
Change authorities receive a concise, source-linked account of the proposed change, its stated impact and its unresolved risks.
Critical infrastructure
Log pattern to incident
Amazon Web Services + Snowflake + OpenAI + ServiceNow
Operators receive a proposed incident supported by related log events, with hypotheses separated from observed evidence.
Automation
Cross-system processes that used to require manual coordination.
These workflows chain enterprise integration, rules and code without requiring AI to be useful.
New employee provisioning
- Trigger
- An HR system creates a new hire record.
- Systems
- HR platform, AD/LDAP, SSO, ITSM and the email and collaboration suite, reached through prebuilt connector actions.
- Mechanism
- A workflow creates accounts, assigns groups from role mappings, opens equipment and access tickets, and confirms each step against the source record before moving on.
- Human control point
- IT or HR approves any access above the standard role template before it is granted.
Outcome: A new starter has working accounts and equipment on day one, with a single record of what was granted and by whom.
See the Enterprise Automation solutionVendor invoice reconciliation
- Trigger
- An invoice arrives in the accounts payable inbox or ERP intake.
- Systems
- ERP, procurement system and a database holding purchase order and receipt records.
- Mechanism
- Rules match the invoice to a purchase order and goods receipt; a code node normalises currency and line items; mismatches are flagged rather than silently adjusted.
- Human control point
- Finance reviews and approves any invoice that does not match automatically, with the discrepancy shown alongside the source records.
Outcome: Matched invoices post without manual handling, and every exception carries the evidence needed to resolve it quickly.
See the Enterprise Automation solutionInfrastructure change through a governed catalog
- Trigger
- An engineer submits a request through the Foundry self-service catalog.
- Systems
- Kubernetes, Terraform, cloud provider APIs and the identity provider for access checks.
- Mechanism
- The request is validated against policy, provisioned through Terraform and Kubernetes actions, and tagged to the requesting team and cost centre.
- Human control point
- Requests above the pre-approved policy threshold route to a named approver before provisioning runs.
Outcome: Engineers get infrastructure without a ticket queue, and platform teams keep a record of who requested what and under which policy.
See the Foundry solutionCross-system customer onboarding
- Trigger
- A signed contract is recorded in the CRM.
- Systems
- CRM, ERP, billing system and the identity provider for portal access provisioning.
- Mechanism
- The workflow creates the customer record across systems, sets up billing terms from the contract fields, and provisions portal access.
- Human control point
- A revenue operations reviewer confirms billing terms before the first invoice is generated.
Outcome: Customer setup across four systems happens as one tracked process instead of four separate manual entries.
See the Enterprise Automation solutionAI
Where a self-hosted model adds judgment rules cannot express.
Every case pairs a model with a defined human control point. None of these describe autonomous decision-making.
Contract clause extraction and review
- Trigger
- A contract is uploaded for review before signature.
- Systems
- Document store, contract management system and a self-hosted model with a legal clause knowledge base.
- Mechanism
- A self-hosted model extracts key clauses and flags deviations from standard terms; deterministic rules check for missing mandatory clauses first.
- Human control point
- Legal reviews flagged deviations and either accepts the extraction or corrects it before the contract proceeds.
Outcome: Reviewers spend time on genuine deviations instead of reading every clause of every contract, with corrections captured for future refinement.
See the Governed AI Automation solutionSupport ticket triage and drafting
- Trigger
- A customer support ticket is created.
- Systems
- Ticketing system, knowledge base and a self-hosted model for classification and draft response generation.
- Mechanism
- The model classifies intent and drafts a response from the retrieved knowledge base articles; rules route billing and legal topics to a human queue automatically.
- Human control point
- An agent reviews and edits the draft before it is sent; nothing goes to a customer unreviewed for the flagged categories.
Outcome: Response drafting time drops for common categories while sensitive topics keep full human authorship.
See the Governed AI Automation solutionSovereign knowledge assistant over internal documentation
- Trigger
- An employee asks a question through an internal chat interface.
- Systems
- Internal document repositories, a self-hosted model and a governed knowledge base built from approved sources.
- Mechanism
- Retrieval pulls relevant passages from the approved knowledge base only, and the model answers with citations back to source documents.
- Human control point
- Access to each knowledge base follows the employee's existing permissions, so the assistant cannot surface documents the person could not already read.
Outcome: Employees get answers sourced from current internal material without documents or queries leaving the organisation's environment.
See the Sovereign AI solutionCredit and limit decision support
- Trigger
- A credit limit increase request is submitted.
- Systems
- ERP, CRM and a self-hosted model summarising account correspondence.
- Mechanism
- Deterministic checks apply policy limits first; a model summarises qualitative signals such as repeated disputes for cases that pass the checks.
- Human control point
- A credit officer approves, adjusts or rejects with a recorded reason, using the rule outcome and model summary together.
Outcome: Clear-cut cases clear automatically, and the officer's time concentrates on the cases that need judgment.
See the Human-in-the-Loop solutionSecurity
Detection, validation and protection with a human in the response path.
Security use cases keep a reviewer between a finding and any consequential action, and describe scope honestly, including what Atlas does not do.
Continuous external attack surface validation
- Trigger
- A scheduled discovery run or a change in DNS and certificate records.
- Systems
- Nexus Atlas discovery and validation engine, asset inventory and ticketing system.
- Mechanism
- Atlas enumerates internet-facing assets, validates exposure and produces a reported finding rather than an unsupervised exploitation attempt.
- Human control point
- Security reviews validated findings before any remediation ticket is raised or exposure is treated as confirmed.
Outcome: Security teams see what changed on their external surface between reviews instead of only during periodic point-in-time scans.
See the External Attack Surface Management solutionAI application red-teaming before release
- Trigger
- A new AI-enabled feature or model endpoint is registered for release.
- Systems
- Nexus Cerberus AI endpoint registry, model registry and reporting pipeline.
- Mechanism
- Cerberus runs OWASP LLM Top 10 test cases mapped to MITRE ATLAS against the registered endpoint and produces signed findings with an AI-BOM of the model's dependencies.
- Human control point
- Application and security owners jointly decide whether findings block release, based on the signed report.
Outcome: AI features go live with a documented security review rather than an assumption that a general model is safe by default.
See the AI Application Security solutionSovereign detection and response
- Trigger
- A detection rule fires on ingested log and telemetry data.
- Systems
- Nexus Sentinel SIEM and SOAR, endpoint and network telemetry sources, ticketing and communication tools.
- Mechanism
- Sentinel correlates the alert against related events, opens a case and can run pre-approved containment actions such as isolating a host.
- Human control point
- An analyst confirms or overrides any containment action before it executes, except for actions explicitly pre-approved for automatic execution.
Outcome: Case creation and initial triage happen immediately on detection, and analysts start from an assembled case rather than a raw alert queue.
See the Security Operations solutionAPI and application traffic protection
- Trigger
- Inbound requests reach a protected application or API.
- Systems
- Nexus Shield, the application or API gateway, and Sentinel for correlated logging.
- Mechanism
- Shield inspects requests against configured rules and blocks or flags traffic matching known attack patterns before it reaches the application.
- Human control point
- Security teams tune and approve rule changes; nothing blocks silently without an entry in the Shield and Sentinel logs.
Outcome: Application traffic is filtered inside the organisation's own boundary rather than through a third-party edge service.
See the Application and API Protection solutionIntegration workflows
Multi-system patterns built from connectors, triggers and code.
Each pattern combines pre-built actions, trigger sources, generic interfaces and Rules + Code in one governed run.
Invoice intake and reconciliation with object storage, SAP and SQL
- Trigger
- New invoice file in AWS S3 or Azure Blob Storage
- Systems
- AWS S3 or Azure Blob Storage, SAP, MSSQL or Oracle
- Mechanism
- A file lands in object storage, SAP and a reconciliation database are checked, Rules + Code matches the result and exceptions go to a named reviewer.
- Human control point
- Finance reviewer approves unmatched or out-of-tolerance invoices
Governed employee access changes with ServiceNow, Okta and Microsoft 365
- Trigger
- New or updated ServiceNow request
- Systems
- ServiceNow, Okta, Microsoft 365 / Calendar
- Mechanism
- A ServiceNow request carries the context, an approver authorises the change, the identity step runs and the outcome returns to the ticket.
- Human control point
- Access owner approves before the identity change runs
Salesforce to SAP order handoff without replacing either system
- Trigger
- Salesforce trigger polling new or updated records via SOQL
- Systems
- Salesforce, SAP
- Mechanism
- A Salesforce record detected by SOQL polling is validated, handed to SAP through the connector or an RFC call, and reconciled back.
- Human control point
- Commercial or finance owner reviews records that fail validation
Customer issue escalation across Salesforce, ServiceNow and Jira
- Trigger
- Salesforce or ServiceNow record change
- Systems
- Salesforce, ServiceNow, Jira
- Mechanism
- One governed run carries an escalation from the customer record to service operations and engineering, and keeps the handoff state visible.
- Human control point
- Service owner confirms escalation into engineering
Event-driven cloud operations with AWS, ArgoCD and Kubernetes events
- Trigger
- AWS CloudWatch, SNS or SQS, ArgoCD or Kubernetes event
- Systems
- AWS CloudWatch, SNS, SQS, ArgoCD, Kubernetes
- Mechanism
- CloudWatch, SNS, SQS, ArgoCD and Kubernetes events start a governed runbook, with approval before any change that policy marks as consequential.
- Human control point
- Operator approves production-affecting remediation
Lead handoff from HubSpot to Salesforce with governed enrichment
- Trigger
- Webhook, HTTP poller or schedule
- Systems
- HubSpot, Salesforce
- Mechanism
- A lead moves from marketing to sales with deduplication, enrichment and routing rules applied in one recorded run.
- Human control point
- Sales operations reviews ambiguous matches before a record is created
Payment and subscription events into CRM and ERP
- Trigger
- Stripe trigger source or scheduled subscription check
- Systems
- Stripe, PayPal Subscriptions, Salesforce, SAP
- Mechanism
- A Stripe event or PayPal subscription change is validated and reflected in Salesforce and SAP under one execution record.
- Human control point
- Finance reviews events that would change revenue records outside policy
Security incident enrichment and ticketing from cloud telemetry
- Trigger
- AWS CloudWatch or CloudWatch Events
- Systems
- AWS CloudWatch, ServiceNow or Jira
- Mechanism
- An AWS CloudWatch event is enriched with Rules + Code and becomes a ServiceNow or Jira work item with the evidence attached.
- Human control point
- Security analyst decides on containment beyond ticket creation
Dedicated integration patterns
- Salesforce + SAP
- Workday + Okta
- ServiceNow + Jira
- HubSpot + Salesforce
- Shopify + NetSuite
- Stripe + Xero
- AWS + ServiceNow
- Microsoft 365 + Okta
- SAP + ServiceNow
- Odoo + SAP
- Workday + SAP
- Coupa + SAP
- Workday + ServiceNow
- Salesforce + ServiceNow
- Salesforce + NetSuite
- HubSpot + SAP
- HubSpot + NetSuite
- Shopify + SAP
- Stripe + SAP
- Stripe + NetSuite
- SAP Concur + NetSuite
- Workday + Salesforce
- Snowflake + SAP
- Snowflake + Salesforce
- PayPal + NetSuite
- Ivanti + ServiceNow
- Shopify + Odoo
- NetSuite + Xero
- ADP + Workday
- OpenAI + Salesforce
Architecture session
Describe your process and we will map it against these patterns.
Most real processes combine two or three of the mechanisms above. A short conversation is enough to sketch the trigger, systems and control points for your specific case.