Sign in Start free trial
IT

IT Operations Management

Know what you run, see what is failing, and fix it with confirmation.

IT operations management keeps an accurate map of your services and items, discovers them from cloud, identity, devices and your own network, and turns monitoring noise into a few explained alert groups. Runbooks fix problems safely: anything that changes a system waits for a person.

64 features available 6 planned 7 capability areas
Highlights

What stands out

Alerts that explain themselves

Each alert group states in words why its alerts belong together, so responders trust it.

Service maps proposed four ways

Mapping proposes members with evidence; a person confirms, so the map stays honest.

Collector built to be safe

Outbound only, signed, read-only and limited to your ranges.

Runbooks with two-person control

Destructive steps need a second person; nothing runs until two people switch runbooks on.

Impact in one click

See what breaks, and which customers feel it, before you change an item.

Capabilities

Every feature, in one list

Search or filter. Planned items are labelled; everything else is available now.

Configuration database

A trustworthy record of what the company runs and how it fits together.

13 features
  • Item classes and attributesDefine classes of configuration items with their attributes, typed relationships and a full history of changes.
  • Identification and reconciliationItems reported by several sources are matched and merged by rule; a more trusted source owns a field, and a merge is always a person's decision.
  • Business and technology service modelModel business services, technology services, service instances and offerings.
  • Health and certificationA health view flags items with no owner, no source, never seen, stale, uncertified or without relationships, and owners certify their items on a cycle.
  • Import and exportBring items in and out as CSV in the service-model shape.
  • Sensitive itemsA sensitive item or class is invisible to anyone without the right to change or approve configuration, in lists, maps, counts, exports, reports and the API.
  • Nothing is deletedItems are retired and relationships ended, so history stays intact.
  • Impact analysisSee what fails when one item does, with a bounded, cycle-safe walk that incident and change screens reuse.
  • Class inheritanceClasses inherit attributes from parent classes, so a server class carries the common item fields.
  • Blocked values and duplicate proposalsA value that would override a more trusted source is filed for a person to decide, and likely duplicates are proposed for merging.
  • Item history and dated relationshipsItem changes are an append-only history and relationships are dated, ended rather than deleted.
  • File feeds never overwrite peopleCSV or JSON feeds report as a low-trust source, so a file never overwrites a person or a more trusted source.
  • Restricted items in a blast radiusImpact shows a sensitive item as Restricted item to people who may not see it.

Discovery from cloud, identity and devices

Cloud accounts, directories and device management feed the database on a schedule.

12 features
  • Cloud and SaaS discoveryRead-only discovery through the provider APIs of cloud platforms and identity directories, with least-privilege permissions. Live account pilots depend on your accounts.
  • Device management feedsBring in managed devices from device management tools.
  • Dry run before runA dry run shows what a run would create, change and report as conflicts without writing anything.
  • Source trust levelsEach source has a trust level so a person's edit beats a feed and a cloud feed beats a file import.
  • Stale item handlingAn item a complete run no longer reports is marked stale after a set number of misses, never deleted; a partial run never marks anything.
  • Run historyEach run keeps counts of items seen, new, changed, conflicts, relationships and stale items.
  • IT actionsFrom IT, create or disable a user, reset multi-factor sign-in, add to a group, assign a licence, lock or wipe a device, or tag or stop a cloud resource; risky actions need a different person to approve.
  • Scheduled runsDiscovery runs off, hourly, daily or weekly per company, and can be simulated outside production.
  • Device match proposalsA device feed updates the register on an exact serial match and proposes uncertain matches for a person to confirm.
  • IT actions in automationsThe automation builder can call IT actions; routine ones run, risky ones become a prepared task awaiting approval.
  • Discovery coverage reportsReports on discovery runs and how much of the estate each source covers.
  • Discovery breadth: mainframe, plant systems, network configurationPlannedDiscover mainframes, operational technology and network device configuration; this is not built and may come through partners.

Network collector

A small program inside your network finds what the cloud feeds cannot see.

9 features
  • Agentless network discoveryChecks which hosts answer, looks up names, reads network devices and reads servers and workstations, all read-only.
  • Outbound-only and signedThe collector opens no inbound port, has its own certificate, and every instruction, probe pack and new build is signed and verified before it runs.
  • Only your rangesScanning is limited to the ranges an administrator allows, and the collector can hold its own allow-list that wins.
  • Sealed loginsNetwork, server and workstation read logins are stored encrypted, never shown, and sent to a collector only encrypted to that collector.
  • Network topologyNeighbour information becomes connects-to relationships in the database.
  • Probe packsPacks for network basics, network devices over SNMP, and servers over SSH and WinRM select probes and hold no code of their own.
  • Collector fleet managementCreate collectors with a single-use token, pause, resume, rotate certificates, revoke, and propose ranges from the networks a collector reports.
  • Signed staged updatesNew collector builds are signed, staged by channel, never downgrade, and roll back if they cannot reach  AI.
  • Scheduled scans and dry runsPrepare, schedule or cancel scans; a dry run changes nothing.

Service mapping and availability

Know which items make up each service and how much of it was lost.

10 features
  • Four mapping methodsEntry-point pattern, tag or label, observed traffic and seen-together each propose members with a confidence and the evidence in words; a person confirms.
  • Map versions and driftEvery change to a service map is a version; compare two or see drift from the last.
  • Service availabilityAvailability against target with every minute of loss explained, planned maintenance agreed by a person and outages reclassified by a person.
  • Customer and money viewOutages show the customers and revenue exposed.
  • Service map screenBrowse the map visually from a service down to its items.
  • Traffic flow importA flow file or collector neighbours feed the observed-traffic mapping method.
  • Customer contracts mapped to servicesMap customer contracts to a service to see revenue per hour and the credit due on an outage.
  • Cloud cost by servicePlannedImport cloud billing, allocate cost by tag, account and service, and show cost per business service.
  • Cloud tag and policy compliancePlannedTag compliance, budget alerts and idle or oversized resource recommendations with an owner and savings.
  • Federated read of external sourcesPlannedRead configuration data in place from external sources and export to a data warehouse without copying.

Events and alerts

Take in alerts from monitoring tools and turn the flood into a short list of explained groups.

15 features
  • Monitoring feedsReceive alerts by signed webhook or pull from supported tools, normalised into one shape. Live accounts depend on your monitoring tools.
  • De-duplicationRepeats of the same alert are counted on one alert; a recovery clears it.
  • Rate and size limitsFeeds that send too fast are slowed, and bodies and batches are capped.
  • Bind alerts to itemsA resource is matched to a configuration item only when exactly one matches; unknown resources go to a not-in-database queue.
  • Correlation with a reasonAlerts group by dependency, service, tag or a recent change, and every group keeps its reason in words.
  • Draft, backtest and approve rulesA new correlation rule is a draft, replayed on past alerts to show the noise it removes, then promoted through approval.
  • Suggested rulesThe system mines history for rules that remove real noise; a person accepts or dismisses.
  • Alert to incidentA policy opens or attaches an incident that names the impacted service and the people it reaches, and pages on-call. It acts through its own limited monitoring identity.
  • Clearing is not resolvingA cleared group leaves the incident alone unless the company chooses to resolve on clear.
  • Alert and group reportsEvent volume, noise reduction, groups with an incident and time to acknowledge.
  • Anomaly detection on logs and metricsPlannedSpot unusual behaviour in logs and metrics; waits on a partner feed.
  • Least-privilege monitoring identityAlert-to-incident acts as a dedicated machine identity that can only open, update and page for its own alert group, never as a person.
  • Rule promotion needs a different approverPromoting a correlation rule to live goes through workflow and the requester cannot approve it.
  • Ordered incident policyAn ordered policy table decides when a group attaches to an open incident, opens one or pages.
  • Wide monitoring coverageAlerts from Datadog, Alertmanager, CloudWatch, Azure Monitor, Grafana and generic webhooks are normalised into one shape.

Runbook automation

Automate fixes safely, with people confirming anything that changes a system.

9 features
  • Fixed step catalogueRunbooks are built from read, diagnose and remediate steps, each with a fixed risk class.
  • Confirm firstRead and diagnose steps run on their own; steps that change something wait for a person; destructive steps need a different person to approve.
  • Off until two people agreeRunbooks are off for every company until an administrator and a second person switch them on.
  • Approved, signed versionsA runbook version is approved through workflow, then signed; an altered version will not run.
  • Dry run and killPlan every step without executing, and stop a running step with a signed instruction.
  • Run on the collectorSteps run on your collector, which checks the instruction again; timeouts apply and output is scrubbed of secrets.
  • Local safeguards on the collectorThe collector rebuilds each command from its own catalogue and re-checks risk and confirmations against its local mode, which is read-only by default.
  • Secrets scrubbed from logsRun logs record who confirmed and approved each step and strip secrets from output.
  • Remote targets and Windows stepsPlannedRun steps on a remote host over SSH and on Windows machines; today steps run on the collector's own host and Windows has no step commands.

On-call and incidents

Alerts reach the right person through the one paging path.

2 features
  • Paging through service managementAlert policies page the on-call schedule in IT service management; there is a single paging path.
  • Pipeline-aware change correlationRecent changes feed correlation so a bad deploy explains the alert storm after it.
Assistant

What  AI prepares here

It prepares the work. A person confirms it.

  • List and explain items, relationships, impact and health findings
  • Read alerts, groups and the reason a group exists
  • Draft a correlation rule from a sentence (not saved)
  • Connect a discovery source as a draft and prepare an IT action as a draft
  • Ask the mapping methods to propose service members and read availability
  • Draft a runbook and plan a dry run
  • Never: merge or certify items, accept a blocked value, confirm a service member, promote a rule, open or attach an incident for a group, run discovery, approve or carry out an action. Those stay with people.
Inside every hub

Workflow, reporting, integration, migration

The same four building blocks, specific to IT Operations Management.

Workflow

Rule promotion, risky IT actions and runbook versions go through the shared workflow module, with the requester or author excluded from approving.

  • Promote a correlation rule to live after a backtest
  • Approve a risky IT action such as a device wipe by a different person
  • Approve a runbook version before it is signed and published
  • Certify items by their owner on a cycle
  • Agree planned maintenance so it does not count against availability

Reporting and analytics

Configuration health and alert reports on the shared reporting module, respecting sensitive items.

Datasets

Configuration items and relationshipsAlertsAlert groupsDiscovery runs

Dashboards

Configuration ownerIT at a glanceAlert noise and time to acknowledge

Integration

Configuration items are readable through the API and written through the data integration feed; monitoring tools post signed alerts or are pulled.

/v1 itom_cis, itom_ci_rels (write through the itom.cis integration feed)

Events

itom.ci.saveditom.ci.retireditom.ci.merged

Connectors

AWSGoogle CloudOktaGoogle WorkspaceIntuneJamfDatadogPrometheusCloudWatchAzure MonitorGrafanaGeneric signed webhook

Migration

Configuration item import by CSV is available; dedicated packs for other configuration databases are planned.

  • CSV import and export of itemsavailable
  • ServiceNow configuration databaseplanned
  • BMC Helix and Remedyplanned
  • Ivantiplanned
Mobile

On the phone

Not a phone-first hub. On-call acknowledgement on the phone app is planned.

Who uses it

IT asset managerService ownerIT managerResolverService desk agentNetwork administratorChange manager
Questions

Frequently asked

Does discovery change anything in our systems?

No. Discovery and collector probes are read-only. IT actions are separate, confirm-first and approved by a different person when risky.

What does the collector need?

A machine in your network that can call  AI over HTTPS; it opens no inbound port.

Can the system delete or merge items by itself?

No. Items are retired, never deleted, and merging is a person's decision.

Can runbooks run by default?

No. Runbooks are off until an administrator and a second person switch them on.

How are alerts grouped?

By dependency, service, tag or recent change, with the reason shown on every group.

Does it detect anomalies in logs?

Not yet; this is planned and depends on a partner feed.

Who can see sensitive items?

Only people allowed to change or approve configuration.

See IT Operations Management on  AI

Talk to us