Enterprise AI workflow library

See what AI can
actually do inside
your business.

Explore practical AI systems and agentic workflows designed around real business processes, from customer support and software engineering to finance, healthcare, manufacturing and operations.

AI assistant

Answers questions and generates information.

AI copilot

Assists a human while they perform work.

AI agent

Reasons, uses tools and executes actions.

Multi-agent system

Specialized agents collaborate across a process.

Find a starting point

Complete workflows,
not isolated demos.

Filter the library by industry, business function or AI type. Every workflow shows the systems involved, human checkpoints, autonomy level and potential value areas.

Showing 40 workflows

Technology & SaaS01
AI Agent

Autonomous Customer Support Resolution

Support teams switch between multiple systems to understand and resolve one customer issue.

Ticket, email or chat arrivesClassify intent and severityRetrieve authorized knowledgeCheck account and product context
AutonomyLevel 4
ReadinessReady Now
Explore workflow
AI handles

Customer context, knowledge, diagnostics and resolution agents

Workflow
  1. Classify intent and severity
  2. Retrieve authorized knowledge
  3. Check account and product context
  4. Recommend or execute resolution
  5. Monitor follow-up and update the knowledge base
Systems involved
ZendeskSalesforceJiraSlackProduct APIs
Human checkpoint

Approve refunds, account changes and other sensitive actions.

Built-in controls
Agent identityPermission-aware retrievalTool permissionsApproval gatesAudit trail
Potential value areas
Faster resolutionFewer repetitive ticketsHigher first-contact resolution
Related CodeCrux capabilities
Technology & SaaS02
Multi-Agent System

Agentic Software Engineering

Engineering teams lose time translating issues into tested, reviewed and deployable changes.

Jira or GitHub issue is createdUnderstand requirementsAnalyze the repositoryCreate an implementation plan
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Requirements, coding, testing and security review agents

Workflow
  1. Understand requirements
  2. Analyze the repository
  3. Create an implementation plan
  4. Write and test code
  5. Open a pull request and monitor CI/CD
Systems involved
GitHubGitLabJiraCI/CDKubernetes
Human checkpoint

Review pull requests and approve protected branch or production changes.

Built-in controls
Repository permissionsProtected filesBranch restrictionsDeployment approvalsCI policy
Potential value areas
Shorter delivery cyclesMore consistent testingHigher developer productivity
Related CodeCrux capabilities
Technology & SaaS03
AI Copilot

SaaS Customer Success Agent

Customer success managers lack one reliable view of product health, support history and renewal risk.

Product health signal or renewal windowCombine product, CRM, support and billing signalsDetect risk or opportunityExplain the account context
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Customer health, risk and outreach agents

Workflow
  1. Combine product, CRM, support and billing signals
  2. Detect risk or opportunity
  3. Explain the account context
  4. Draft an intervention
  5. Monitor the customer response
Systems involved
CRMProduct analyticsSupport platformBillingEmail
Human checkpoint

Review high-value outreach, retention offers and escalation decisions.

Built-in controls
Customer data permissionsCRM write accessOutreach approvalTraceable recommendations
Potential value areas
Earlier churn detectionBetter renewal preparationMore relevant customer engagement
Related CodeCrux capabilities
Technology & SaaS04
Multi-Agent System

Product Intelligence Agent

Product feedback is fragmented across tickets, reviews, analytics, sales calls and feature requests.

New feedback or scheduled intelligence reviewCollect approved feedback sourcesCluster recurring problemsIdentify emerging trends
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Feedback clustering, trend and impact agents

Workflow
  1. Collect approved feedback sources
  2. Cluster recurring problems
  3. Identify emerging trends
  4. Estimate opportunity impact
  5. Draft a product requirement for review
Systems involved
Support platformCRMAnalyticsApp reviewsProduct backlog
Human checkpoint

Validate themes, prioritize opportunities and approve product requirements.

Built-in controls
Source permissionsEvidence linksHuman prioritizationBacklog approval
Potential value areas
Faster product discoveryClearer customer signalBetter roadmap decisions
Related CodeCrux capabilities
Banking & Financial Services05
Agentic Workflow

Intelligent Loan Processing

Loan teams manually gather, verify and compare documents before a credit decision can be reviewed.

Loan application is submittedCollect and classify documentsExtract financial informationRun KYC and identity checks
AutonomyLevel 3
ReadinessAdvanced
Explore workflow
AI handles

Document, identity, risk and policy agents

Workflow
  1. Collect and classify documents
  2. Extract financial information
  3. Run KYC and identity checks
  4. Evaluate policy and risk
  5. Prepare an underwriting recommendation
Systems involved
Loan originationKYC providerDocument storeCredit dataCRM
Human checkpoint

Make or approve regulated and high-risk underwriting decisions.

Built-in controls
PII protectionSeparation of dutiesPolicy versioningApproval workflowEvidence archive
Potential value areas
Shorter processing timeFewer manual handoffsMore consistent review packages
Related CodeCrux capabilities
Banking & Financial Services06
AI Agent

Fraud Investigation Agent

Fraud analysts spend hours collecting context from transactions, devices, locations and account history.

Transaction alert is createdGather customer and transaction historyAnalyze device and location contextFind related accounts
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Investigation, entity and risk analysis agents

Workflow
  1. Gather customer and transaction history
  2. Analyze device and location context
  3. Find related accounts
  4. Score risk and compile evidence
  5. Prepare a case for analyst review
Systems involved
Core bankingFraud platformCustomer profileDevice intelligenceCase management
Human checkpoint

Review evidence, decide escalation and take account action.

Built-in controls
Least privilegeCase access controlsExplainable evidenceAnalyst approvalAudit trail
Potential value areas
Faster investigationsBetter analyst focusMore complete case evidence
Related CodeCrux capabilities
Banking & Financial Services07
AI Copilot

Financial Advisor Copilot

Advisors need to synthesize client profiles, portfolios, market research and risk information before meetings.

Client review or meeting preparationRetrieve the client profileAnalyze portfolio and risk profileResearch approved market sources
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Research, scenario and presentation agents

Workflow
  1. Retrieve the client profile
  2. Analyze portfolio and risk profile
  3. Research approved market sources
  4. Generate scenarios and recommendations
  5. Prepare an advisor-reviewed presentation
Systems involved
CRMPortfolio systemMarket dataResearch libraryDocument platform
Human checkpoint

Own regulated advice, recommendations and all client-facing decisions.

Built-in controls
Approved source listRole-based accessDisclosure checksAdvisor reviewVersioned outputs
Potential value areas
Better meeting preparationFaster researchMore consistent client communication
Related CodeCrux capabilities
Banking & Financial Services08
Multi-Agent System

AML Investigation Workflow

AML investigators manually resolve entities and assemble evidence across transaction and research systems.

AML alert is raisedResolve related entitiesAnalyze transaction relationshipsRetrieve historical investigations
AutonomyLevel 3
ReadinessAdvanced
Explore workflow
AI handles

Entity resolution, transaction graph and evidence agents

Workflow
  1. Resolve related entities
  2. Analyze transaction relationships
  3. Retrieve historical investigations
  4. Search approved adverse information
  5. Compile evidence for investigator review
Systems involved
Transaction platformCase managementKYC dataResearch sourcesEntity graph
Human checkpoint

Review the case, apply judgment and decide escalation or reporting.

Built-in controls
Source authorizationEvidence provenanceInvestigator approvalRetention policyAuditability
Potential value areas
More complete investigationsReduced research effortFaster case preparation
Related CodeCrux capabilities
Healthcare & Life Sciences09
AI Copilot

Clinical Documentation Copilot

Clinicians spend valuable time turning patient interactions into structured notes and coding suggestions.

Patient interaction is completeTranscribe the interactionExtract clinical informationApply medical context
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Transcription, clinical extraction and documentation agents

Workflow
  1. Transcribe the interaction
  2. Extract clinical information
  3. Apply medical context
  4. Draft a note and coding suggestions
  5. Present the draft for clinician review
Systems involved
EHRClinical documentationSpeech platformCoding systemIdentity provider
Human checkpoint

Review, edit and approve all clinical notes and coding decisions.

Built-in controls
Patient data accessClinician identityNo autonomous diagnosisReview gateAudit log
Potential value areas
Less documentation burdenFaster note completionMore consistent records
Related CodeCrux capabilities
Healthcare & Life Sciences10
AI Agent

Patient Service Agent

Patients need answers across appointments, records, billing and common questions, often across disconnected systems.

Patient request arrivesVerify identityUnderstand the requestRetrieve the authorized record
AutonomyLevel 4
ReadinessReady Now
Explore workflow
AI handles

Identity, intent and service agents

Workflow
  1. Verify identity
  2. Understand the request
  3. Retrieve the authorized record
  4. Resolve or route the request
  5. Escalate when a human is needed
Systems involved
Patient portalSchedulingBillingEHRKnowledge base
Human checkpoint

Handle clinical, sensitive, ambiguous and exception cases.

Built-in controls
Identity verificationMinimum necessary accessSensitive-topic escalationAudit trail
Potential value areas
Faster patient responseLower service workloadMore consistent support
Related CodeCrux capabilities
Healthcare & Life Sciences11
Agentic Workflow

Prior Authorization Intelligence Workflow

Prior authorization teams spend time collecting evidence, interpreting policy and tracking submissions.

Treatment authorization is requestedRetrieve the patient recordAnalyze payer requirementsCollect required evidence
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Record retrieval, policy and documentation agents

Workflow
  1. Retrieve the patient record
  2. Analyze payer requirements
  3. Collect required evidence
  4. Prepare submission documents
  5. Monitor status and draft an appeal when needed
Systems involved
EHRPayer portalPolicy libraryDocument storeCase management
Human checkpoint

Review clinical documentation, submission accuracy and appeals.

Built-in controls
Patient consentPayer policy versioningHuman reviewDocument audit trail
Potential value areas
Faster submission preparationFewer missing documentsBetter status visibility
Related CodeCrux capabilities
Healthcare & Life Sciences12
Multi-Agent System

Life Sciences Research Agent

Researchers need to connect literature, trials, internal findings and structured scientific knowledge.

Scientific question is submittedSearch approved literatureRetrieve relevant trial dataConnect internal research
AutonomyLevel 2
ReadinessAdvanced
Explore workflow
AI handles

Literature, trial, knowledge graph and evidence agents

Workflow
  1. Search approved literature
  2. Retrieve relevant trial data
  3. Connect internal research
  4. Evaluate evidence quality
  5. Generate hypotheses for researcher review
Systems involved
Research repositoryClinical trial dataKnowledge graphDocument storeApproved external sources
Human checkpoint

Evaluate findings, validate hypotheses and make scientific decisions.

Built-in controls
Approved sourcesCitation provenanceResearcher accessEvidence scoringReview workflow
Potential value areas
Faster research synthesisMore connected knowledgeBetter evidence discovery
Related CodeCrux capabilities
Insurance13
AI Agent

Claims Processing Agent

Claims teams review documents, images, policies and fraud signals across multiple systems.

Claim is submittedAnalyze claim documents and imagesRetrieve policy coverageAssess damage or loss context
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Document, policy, damage and fraud signal agents

Workflow
  1. Analyze claim documents and images
  2. Retrieve policy coverage
  3. Assess damage or loss context
  4. Identify fraud signals
  5. Prepare a settlement recommendation
Systems involved
Claims platformPolicy systemImage storeCustomer profileCase management
Human checkpoint

Review complex claims, exceptions, fraud signals and settlement decisions.

Built-in controls
Claim access policyEvidence linksAdjuster approvalFraud escalationAudit trail
Potential value areas
Faster claim triageMore consistent reviewReduced manual effort
Related CodeCrux capabilities
Insurance14
AI Copilot

Underwriting Copilot

Underwriters need a clear risk summary across applications, historical data and policy rules.

Insurance application arrivesExtract application dataEnrich approved risk signalsAnalyze historical patterns
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Enrichment, risk and policy analysis agents

Workflow
  1. Extract application data
  2. Enrich approved risk signals
  3. Analyze historical patterns
  4. Apply policy rules
  5. Prepare a coverage and pricing recommendation
Systems involved
Policy administrationUnderwriting workbenchRisk dataCRMDocument store
Human checkpoint

Own underwriting judgment, pricing and coverage approval.

Built-in controls
Data lineageRule versioningUnderwriter reviewAccess controls
Potential value areas
Faster application reviewClearer risk contextMore consistent underwriting packages
Related CodeCrux capabilities
Insurance15
AI Agent

Insurance Customer Service Agent

Policyholders need quick answers and simple policy actions without navigating complex policy language.

Customer request arrivesVerify the policyholderUnderstand the requestRetrieve coverage details
AutonomyLevel 4
ReadinessReady Now
Explore workflow
AI handles

Policy, account and action agents

Workflow
  1. Verify the policyholder
  2. Understand the request
  3. Retrieve coverage details
  4. Prepare or execute an allowed action
  5. Escalate exceptions
Systems involved
Policy systemCustomer portalClaims platformCRMContact center
Human checkpoint

Handle disputes, exceptions, sensitive changes and coverage interpretation.

Built-in controls
Identity verificationPolicy permissionsAction allowlistHuman escalationAudit trail
Potential value areas
Faster answersLower service workloadBetter policyholder experience
Related CodeCrux capabilities
Insurance16
Agentic Workflow

Renewal Intelligence Agent

Renewal teams need to understand changing risk, policy history and customer behavior before outreach.

Policy enters renewal windowReview claims and policy historyAnalyze customer behaviorDetect changes in risk
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Policy history, risk and outreach agents

Workflow
  1. Review claims and policy history
  2. Analyze customer behavior
  3. Detect changes in risk
  4. Draft a renewal recommendation
  5. Prepare personalized outreach for review
Systems involved
Policy systemClaims platformCRMCustomer analyticsEmail
Human checkpoint

Review renewal recommendations, pricing and customer communication.

Built-in controls
Customer permissionsPricing approvalOutreach reviewRecommendation evidence
Potential value areas
Better renewal preparationMore relevant offersEarlier risk visibility
Related CodeCrux capabilities
Retail & E-commerce17
AI Agent

Shopping Agent

Customers need help translating intent into product choices, comparisons and an approved cart.

Customer expresses a shopping intentUnderstand preferencesSearch authorized product dataCompare recommendations
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Preference, product and inventory agents

Workflow
  1. Understand preferences
  2. Search authorized product data
  3. Compare recommendations
  4. Check inventory
  5. Prepare a cart for customer approval
Systems involved
Commerce platformProduct catalogInventoryPricingCart
Human checkpoint

Customer approves the cart and completes checkout.

Built-in controls
Catalog permissionsPrice accuracyCustomer approvalPurchase boundary
Potential value areas
More relevant discoveryFaster product comparisonHigher shopping convenience
Related CodeCrux capabilities
Retail & E-commerce18
AI Agent

Customer Service & Returns Agent

Returns and order questions require repetitive lookup across orders, policies, risk checks and fulfillment systems.

Customer asks about an order or returnVerify the customerLook up order contextInterpret return policy
AutonomyLevel 4
ReadinessReady Now
Explore workflow
AI handles

Order, policy and risk agents

Workflow
  1. Verify the customer
  2. Look up order context
  3. Interpret return policy
  4. Check fraud and risk signals
  5. Execute an allowed action or escalate
Systems involved
Commerce platformOrder managementReturns platformPaymentsCRM
Human checkpoint

Approve exceptions, high-value refunds and disputed cases.

Built-in controls
Identity verificationRefund limitsRisk checksApproval gatesAudit trail
Potential value areas
Faster responseLower service workloadConsistent policy handling
Related CodeCrux capabilities
Retail & E-commerce19
Agentic Workflow

Inventory Optimization System

Merchandising and planning teams balance demand, promotions, supply and inventory across disconnected signals.

Scheduled forecast or supply disruptionCombine demand and promotion signalsForecast demandDetect inventory risk
AutonomyLevel 3
ReadinessAdvanced
Explore workflow
AI handles

Demand forecast, inventory and supplier agents

Workflow
  1. Combine demand and promotion signals
  2. Forecast demand
  3. Detect inventory risk
  4. Recommend replenishment
  5. Coordinate suppliers after planner approval
Systems involved
ERPInventory platformSales dataSupplier systemsAnalytics
Human checkpoint

Approve replenishment plans and supplier commitments.

Built-in controls
Planner permissionsScenario evidenceERP write approvalSupplier access policy
Potential value areas
Better inventory visibilityFewer stockout surprisesFaster planning cycles
Related CodeCrux capabilities
Retail & E-commerce20
Multi-Agent System

Merchandising Intelligence Agent

Merchandisers need to connect customer behavior, search trends, margins, inventory and competitors.

Weekly category review or market signalAnalyze behavior and search trendsReview inventory and marginFind product opportunities
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Trend, pricing and promotion agents

Workflow
  1. Analyze behavior and search trends
  2. Review inventory and margin
  3. Find product opportunities
  4. Recommend pricing and promotions
  5. Prepare a review package
Systems involved
Commerce analyticsSearch analyticsInventoryPricingCompetitor data
Human checkpoint

Approve pricing, promotions and merchandising decisions.

Built-in controls
Data permissionsMargin thresholdsApproval workflowEvidence links
Potential value areas
Faster category reviewsBetter opportunity discoveryMore informed promotions
Related CodeCrux capabilities
Manufacturing21
AI Agent

Predictive Maintenance Agent

Maintenance teams need to interpret telemetry, asset history, failure signals and parts availability together.

Machine anomaly is detectedAnalyze telemetryRetrieve maintenance historyPredict failure risk
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Anomaly, asset context and maintenance agents

Workflow
  1. Analyze telemetry
  2. Retrieve maintenance history
  3. Predict failure risk
  4. Check parts availability
  5. Generate a work order recommendation
Systems involved
SCADAIoT platformCMMSERPParts inventory
Human checkpoint

Approve maintenance work and technician action.

Built-in controls
Asset accessOperational boundariesWork order approvalSafety escalationAudit trail
Potential value areas
Less unplanned downtimeFaster diagnosisBetter maintenance planning
Related CodeCrux capabilities
Manufacturing22
Agentic Workflow

Production Optimization Agent

Production planners balance demand, capacity, materials and machine availability under changing constraints.

Planning cycle or constraint changeGather demand and capacityCheck materials and machine availabilityRun schedule scenarios
AutonomyLevel 2
ReadinessAdvanced
Explore workflow
AI handles

Capacity, schedule and constraint agents

Workflow
  1. Gather demand and capacity
  2. Check materials and machine availability
  3. Run schedule scenarios
  4. Explain constraints
  5. Prepare a production plan for approval
Systems involved
ERPMESSupply planningInventoryScheduling
Human checkpoint

Approve the production plan and operational trade-offs.

Built-in controls
Planner accessScenario traceabilityERP/MES approvalConstraint guardrails
Potential value areas
Faster planningMore transparent trade-offsBetter use of capacity
Related CodeCrux capabilities
Manufacturing23
Multi-Agent System

Quality Investigation Agent

Quality engineers investigate defects across images, sensors, batches, machine settings and suppliers.

Defect is detectedAnalyze defect evidenceRetrieve batch and machine historyCompare supplier context
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Vision, batch, machine and root-cause agents

Workflow
  1. Analyze defect evidence
  2. Retrieve batch and machine history
  3. Compare supplier context
  4. Investigate likely root causes
  5. Recommend corrective action
Systems involved
Quality systemMESMachine telemetrySupplier portalDocument store
Human checkpoint

Validate root cause and approve corrective action.

Built-in controls
Quality data permissionsEvidence provenanceEngineer reviewCorrective action approval
Potential value areas
Faster root-cause analysisMore complete evidenceReduced investigation effort
Related CodeCrux capabilities
Manufacturing24
AI Assistant

Engineering Knowledge Agent

Engineers search specifications, manuals, CAD metadata, maintenance reports and historical issues separately.

Engineer asks a technical questionUnderstand the questionSearch authorized engineering sourcesConnect related evidence
AutonomyLevel 1
ReadinessReady Now
Explore workflow
AI handles

Retrieval, evidence and recommendation agents

Workflow
  1. Understand the question
  2. Search authorized engineering sources
  3. Connect related evidence
  4. Answer with citations
  5. Suggest a next action
Systems involved
PLMDocument storeMaintenance systemKnowledge graphIssue tracker
Human checkpoint

Validate technical conclusions and own engineering decisions.

Built-in controls
Permission-aware retrievalSource citationsDocument versioningEngineer review
Potential value areas
Faster issue resolutionLess repeated searchingBetter reuse of engineering knowledge
Related CodeCrux capabilities
Telecommunications & Media25
AI Agent

Network Operations Agent

Network operations teams investigate alerts across topology, telemetry and historical incidents before taking action.

Network alert is raisedMap affected dependenciesInvestigate telemetrySearch historical incidents
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Topology, telemetry and remediation agents

Workflow
  1. Map affected dependencies
  2. Investigate telemetry
  3. Search historical incidents
  4. Recommend remediation
  5. Verify the result after approved action
Systems involved
Network managementObservabilityTopology graphIncident platformChange management
Human checkpoint

Approve changes that affect production services or customer traffic.

Built-in controls
Network identityChange policyProduction approvalRollback planAudit trail
Potential value areas
Faster incident triageReduced investigation effortMore consistent remediation
Related CodeCrux capabilities
Telecommunications & Media26
AI Agent

Customer Retention Agent

Retention teams need to connect usage, billing, support and network quality to understand churn risk.

Churn signal is detectedAnalyze usage and billingReview support and service qualityExplain likely churn reasons
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Health, reason and outreach agents

Workflow
  1. Analyze usage and billing
  2. Review support and service quality
  3. Explain likely churn reasons
  4. Recommend retention action
  5. Prepare customer outreach
Systems involved
CRMBillingNetwork analyticsSupport platformCampaigns
Human checkpoint

Approve retention offers and high-value customer outreach.

Built-in controls
Customer permissionsOffer thresholdsCRM write policyOutreach review
Potential value areas
Earlier churn visibilityMore relevant interventionsFaster account preparation
Related CodeCrux capabilities
Telecommunications & Media27
Agentic Workflow

Field Service Agent

Field teams coordinate diagnosis, equipment, parts, technicians and routes across several operational systems.

Service incident is createdDiagnose remotelyIdentify equipment and locationDetermine parts and skills
AutonomyLevel 3
ReadinessAdvanced
Explore workflow
AI handles

Diagnosis, parts, dispatch and field assistance agents

Workflow
  1. Diagnose remotely
  2. Identify equipment and location
  3. Determine parts and skills
  4. Assign a technician and route
  5. Assist repair and close the incident
Systems involved
Service managementAsset registryInventoryMapsWorkforce scheduling
Human checkpoint

Approve dispatch exceptions and perform physical repair work.

Built-in controls
Technician identityDispatch policySafety checksWork order approvalCompletion evidence
Potential value areas
Faster dispatchBetter first-time fix ratesLess coordination effort
Related CodeCrux capabilities
Telecommunications & Media28
Multi-Agent System

Media Content Intelligence

Media teams need to connect content libraries, audience signals and trends before planning programming or campaigns.

New content signal or planning cycleAnalyze the content librarySegment audiencesIdentify trends
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Audience, trend and content recommendation agents

Workflow
  1. Analyze the content library
  2. Segment audiences
  3. Identify trends
  4. Recommend content and campaigns
  5. Prepare an editorial review package
Systems involved
Content libraryAudience analyticsCampaign platformTrendsCMS
Human checkpoint

Own editorial judgment, programming and campaign approval.

Built-in controls
Content permissionsSource evidenceEditorial reviewPublishing approval
Potential value areas
Faster content planningBetter audience understandingMore informed programming
Related CodeCrux capabilities
Transportation & Logistics29
Multi-Agent System

Logistics Control Tower Agent

Operations teams need to re-plan orders, vehicles, warehouses, traffic and customer ETAs when conditions change.

Disruption or control-tower signalDetect disruptionEvaluate alternative routes and resourcesRe-plan delivery
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Disruption, route and customer communication agents

Workflow
  1. Detect disruption
  2. Evaluate alternative routes and resources
  3. Re-plan delivery
  4. Request operational approval
  5. Update customer ETA and monitor
Systems involved
TMSWMSFleet platformMapsCustomer portal
Human checkpoint

Approve material route, cost and service-level trade-offs.

Built-in controls
Dispatcher identityRoute boundariesApproval policyETA evidenceAudit trail
Potential value areas
Faster disruption responseBetter delivery visibilityFewer manual handoffs
Related CodeCrux capabilities
Transportation & Logistics30
AI Agent

Shipment Exception Agent

Teams investigate delays, damage and missing shipments across carriers, warehouses and route data.

Shipment exception is detectedInvestigate shipment statusCombine carrier and warehouse dataAnalyze the route
AutonomyLevel 4
ReadinessReady Now
Explore workflow
AI handles

Shipment, carrier and resolution agents

Workflow
  1. Investigate shipment status
  2. Combine carrier and warehouse data
  3. Analyze the route
  4. Recommend resolution
  5. Communicate with the customer and open a claim if needed
Systems involved
TMSWMSCarrier APIsCRMClaims platform
Human checkpoint

Approve compensation, claims and exception handling.

Built-in controls
Shipment permissionsCompensation limitsCustomer communication reviewClaims audit
Potential value areas
Faster exception resolutionBetter customer updatesReduced coordination effort
Related CodeCrux capabilities
Transportation & Logistics31
AI Agent

Fleet Maintenance Agent

Fleet managers need to combine vehicle telemetry, maintenance history, parts and schedules.

Vehicle health signal is detectedAnalyze vehicle healthRetrieve maintenance historyPredict failure risk
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Health, maintenance and scheduling agents

Workflow
  1. Analyze vehicle health
  2. Retrieve maintenance history
  3. Predict failure risk
  4. Check parts and service capacity
  5. Prepare a maintenance schedule
Systems involved
Fleet platformTelematicsCMMSParts inventoryScheduling
Human checkpoint

Approve maintenance scheduling and safety-related actions.

Built-in controls
Fleet identitySafety thresholdsWork order approvalMaintenance history
Potential value areas
Less unplanned downtimeBetter fleet availabilityFaster maintenance planning
Related CodeCrux capabilities
Transportation & Logistics32
Agentic Workflow

Intelligent Dispatch Agent

Dispatchers balance driver availability, capacity, traffic, priority and changing delivery requests.

New delivery request or route changeCheck driver and vehicle availabilityEvaluate capacity and priorityOptimize the route
AutonomyLevel 4
ReadinessReady Now
Explore workflow
AI handles

Capacity, routing and dispatch agents

Workflow
  1. Check driver and vehicle availability
  2. Evaluate capacity and priority
  3. Optimize the route
  4. Recommend an assignment
  5. Continuously re-optimize after approval
Systems involved
Dispatch platformFleet systemMapsOrdersDriver app
Human checkpoint

Approve assignments that violate normal policy or service boundaries.

Built-in controls
Driver identityRoute constraintsDispatch approvalCustomer visibility
Potential value areas
More efficient dispatchFaster response to changeBetter vehicle utilization
Related CodeCrux capabilities
Energy & Utilities33
AI Agent

Grid Operations Intelligence

Grid operators need to interpret telemetry, demand, supply, weather and operational risk together.

Grid anomaly or demand signalDetect and classify anomalyAnalyze demand and supplyAdd weather context
AutonomyLevel 2
ReadinessAdvanced
Explore workflow
AI handles

Anomaly, forecast and operations agents

Workflow
  1. Detect and classify anomaly
  2. Analyze demand and supply
  3. Add weather context
  4. Assess operational risk
  5. Prepare an operator recommendation
Systems involved
SCADAGrid platformWeather dataForecastingOperations dashboard
Human checkpoint

Approve grid actions and retain operational accountability.

Built-in controls
Operator identitySafety boundariesApproval gatesAction verificationAudit trail
Potential value areas
Faster situation awarenessBetter operational decisionsReduced analysis effort
Related CodeCrux capabilities
Energy & Utilities34
Agentic Workflow

Field Maintenance Agent

Field maintenance teams coordinate asset history, diagnostics, parts, skills and work orders manually.

Asset alert is receivedRetrieve asset historyRun remote diagnosticsIdentify skills and parts
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Asset, diagnostic, parts and workforce agents

Workflow
  1. Retrieve asset history
  2. Run remote diagnostics
  3. Identify skills and parts
  4. Create a work order
  5. Assign and assist the technician
Systems involved
Asset managementIoT platformCMMSInventoryWorkforce scheduling
Human checkpoint

Approve work, perform repair and close the maintenance record.

Built-in controls
Asset permissionsSafety checksWork order approvalTechnician identityCompletion audit
Potential value areas
Faster maintenance coordinationBetter field preparationMore complete records
Related CodeCrux capabilities
Energy & Utilities35
Multi-Agent System

Energy Forecasting Agent

Energy teams need reliable scenarios across consumption, weather, historical demand and market signals.

Forecasting cycle or market changeCombine demand and weather dataGenerate forecast scenariosAnalyze market signals
AutonomyLevel 2
ReadinessAdvanced
Explore workflow
AI handles

Consumption, weather and scenario agents

Workflow
  1. Combine demand and weather data
  2. Generate forecast scenarios
  3. Analyze market signals
  4. Explain uncertainty
  5. Prepare recommendations for review
Systems involved
MeteringWeather dataMarket dataForecast platformAnalytics
Human checkpoint

Review scenarios and make operational or commercial decisions.

Built-in controls
Data lineageScenario versioningAnalyst reviewApproved sources
Potential value areas
Faster scenario analysisBetter planning visibilityMore informed supply decisions
Related CodeCrux capabilities
Energy & Utilities36
AI Copilot

Customer Energy Advisor

Customers need clear guidance from usage patterns, costs, tariffs and efficiency opportunities.

Customer asks about energy usage or costAnalyze usage patternsExplain cost driversCompare approved tariffs
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Usage, tariff and recommendation agents

Workflow
  1. Analyze usage patterns
  2. Explain cost drivers
  3. Compare approved tariffs
  4. Identify efficiency opportunities
  5. Present personalized recommendations
Systems involved
MeteringBillingTariff systemCRMCustomer portal
Human checkpoint

Customer chooses actions and a representative handles exceptions.

Built-in controls
Customer identityBilling permissionsTariff versioningRecommendation evidence
Potential value areas
Clearer customer communicationFaster supportMore actionable advice
Related CodeCrux capabilities
Professional Services37
Multi-Agent System

Research & Knowledge Agent

Knowledge teams search internal documents, past projects and approved external sources to answer business questions.

Business question is submittedUnderstand the questionSearch approved internal knowledgeRetrieve past project evidence
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Research, retrieval and evidence agents

Workflow
  1. Understand the question
  2. Search approved internal knowledge
  3. Retrieve past project evidence
  4. Evaluate sources
  5. Synthesize a reviewed deliverable
Systems involved
Document storeProject repositoryKnowledge baseApproved researchCollaboration tools
Human checkpoint

Review conclusions, apply expertise and approve the deliverable.

Built-in controls
Permission-aware retrievalSource citationsExpert reviewExternal source policy
Potential value areas
Faster researchMore reusable knowledgeHigher-quality first drafts
Related CodeCrux capabilities
Professional Services38
Agentic Workflow

Proposal & RFP Agent

Proposal teams manually extract requirements and search past work before drafting a compliant response.

RFP is receivedExtract requirementsFind relevant experienceDraft the solution response
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Requirement, experience, solution and compliance agents

Workflow
  1. Extract requirements
  2. Find relevant experience
  3. Draft the solution response
  4. Prepare pricing inputs
  5. Run compliance checks before human review
Systems involved
RFP inboxCRMProposal libraryPricingDocument collaboration
Human checkpoint

Approve solution, pricing, commitments and final submission.

Built-in controls
Approved experience sourcesPricing accessCompliance checksFinal approval
Potential value areas
Faster proposal preparationMore consistent responsesLess repetitive research
Related CodeCrux capabilities
Professional Services39
AI Workflow

Client Delivery Agent

Client delivery teams lose commitments and context between meetings, research, documents and project tracking.

Client meeting or delivery milestoneExtract tasks and commitmentsResearch the next questionsDraft delivery documents
AutonomyLevel 3
ReadinessReady Now
Explore workflow
AI handles

Meeting, task, research and risk agents

Workflow
  1. Extract tasks and commitments
  2. Research the next questions
  3. Draft delivery documents
  4. Update project tracking
  5. Detect risks and prepare a client update
Systems involved
Meeting platformProject managementDocument storeCRMCollaboration tools
Human checkpoint

Review client-facing documents, decisions and risk escalations.

Built-in controls
Client data boundariesDocument reviewProject write permissionsApproval workflow
Potential value areas
Fewer missed commitmentsFaster follow-upBetter delivery visibility
Related CodeCrux capabilities
Professional Services40
AI Agent

Resource & Staffing Intelligence

Staffing teams balance pipeline, skills, availability and project requirements across disconnected systems.

New project requirement or staffing reviewExtract project requirementsMatch approved skills and availabilityGenerate staffing scenarios
AutonomyLevel 2
ReadinessReady Now
Explore workflow
AI handles

Matching, scenario and utilization agents

Workflow
  1. Extract project requirements
  2. Match approved skills and availability
  3. Generate staffing scenarios
  4. Analyze utilization
  5. Prepare a recommendation for manager approval
Systems involved
CRMHRISResource managementSkills directoryProject platform
Human checkpoint

Approve staffing, assignment and people-related decisions.

Built-in controls
Employee data permissionsFairness reviewManager approvalScheduling boundaries
Potential value areas
Faster staffing decisionsBetter utilization visibilityMore transparent scenarios
Related CodeCrux capabilities

Controlled adoption

Autonomy should be earned.

Not every process needs a fully autonomous agent. Start at the level that fits the risk, data and trust required, then scale what works.

01AssistAI provides information.
02RecommendAI proposes the next action.
03Act with approvalAI acts after human approval.
04Within policyAI acts inside permissions.
05Multi-agentHumans supervise exceptions.

One workflow, many specialists

When one agent
isn't enough.

Complex enterprise processes work best when specialized agents collaborate inside clear permissions and a visible orchestration layer.

Enterprise systemsCRMSupportERPKnowledgeIdentity
ORCHESTRATED CUSTOMER OPERATIONS
CustomerAccountKnowledgeDiagnosticAction
Policy controls / human supervisor / audit trail
Control layerPolicyApprovalAuditObservability

Make it specific

What workflow would
you give to an AI agent?

Tell us how your team performs the process today. We will help identify what AI should automate, what stays human-controlled, which systems need integration and how to measure the outcome.

Design my AI workflow