Skip to content
Blueprint · Telecom

Plan the change. Check the result.

A worked example: a parameter change across a cluster of radio sites, planned against live network state and vendor rules, run in an approved window and closed only when service data confirms it, or an approved rollback restores the network.

Telecom
The question

Can we roll out the new handover parameters to cluster 14 tonight, and how will we know it worked?

Inputs

What the agents read.

Change request
Target, scope, window and rollback plan from ITSM
Network state
Inventory and current configuration from OSS and NMS
Performance and alarms
KPIs and alarms before and after the change
Vendor rules
Features supported by each software version
Change policy
Who approves, when to run and when to recover
Run

What happens, step by step. Each step is logged with its sources.

  1. 01

    Read what was approved

    Start with the target, scope, maintenance window and rollback plan.

  2. 02

    Build the network picture

    Pull in current state, recent changes, vendor rules, licences and service health.

  3. 03

    Test the plan

    Look for missing dependencies, unsupported combinations and policy exceptions.

  4. 04

    Prepare the change

    Create target-specific commands, the schedule and clear recovery steps.

  5. 05

    Watch the result

    Compare KPIs and alarms after the change with the agreed baseline.

  6. 06

    Finish or roll back

    Keep a verified result, or run the approved rollback and check again.

Agent Hive · Run 2207
Can we roll out the new handover parameters to cluster 14 tonight?
  1. ReadCHG-2207 scope, window, rollback0.4s
  2. Queriedoss.inventory · 22 cells1.3s
  3. Checkedvendor rules for software 21.40.8s
  4. Flaggedsite 14-10 on 20.9, no licence0.5s
  5. Preparedcommands for 20 cells + rollback0.9s
Answer4 sources
20 of 22cells
Ready for tonight’s window

Proceed on 20 cells. Site 14-10 runs software 20.9 without the handover licence; it needs a specialist before it joins the change.

itsm.changesoss.inventoryvendor rules 21.4licence register
Ask a follow-up…
Approvals

Where people decide.

  • Needs you

    Before execution

    The accountable engineer can change, approve or reject the proposed action.

  • Needs you

    When something does not fit

    Missing prerequisites and unsupported combinations go to a specialist.

  • Needs you

    When evidence is thin

    Your team sets the point where uncertainty stops the workflow and asks for help.

  • Needs you

    When service gets worse

    Service limits and change policy decide when, and by whom, the change is reversed.

Outputs

What you get back.

Verified change record
Plan, approvals, commands and before/after KPIs
Rollback record
Trigger, approver and the network state after recovery
Template proposal
A reusable plan, for specialists to accept or reject
Data sources

Start with the evidence your team already trusts.

The request, network state, vendor rules and service data stay in their systems. Agent Hive connects the pieces that matter for this change.

Fig. 1 · Connected sources
Sources stay in place; Agent Hive reads them through read-only connectors and builds one shared model. Logos are trademarks of their respective owners.
  • ITSM

    The request, scope, window and approvals.

  • OSS, NMS and performance

    Inventory, configuration, KPIs and alarms.

  • Vendor rules and policy

    Supported features by version, licences and change policy.

Agent workflow

A change is not finished when the command runs.

The run closes only after service data confirms the result, or an approved rollback restores the network.

Fig. 2 · Workflow runIllustrative data
The run as the platform draws it: tool steps, specialist agents, joins, branches and the approval it is waiting on. Drag to pan; Ctrl or ⌘ + wheel to zoom.
Knowledge model

The network model: what depends on what.

Sites, cells, software versions, features and licences, joined to the vendor rules and change policies that govern them.

Fig. 3 · Knowledge modelIllustrative data
Cluster 14 in the network model: sites, cells and their neighbours, software versions, the handover feature and licence, and the policies that apply.
Blast radius

Changes ripple. The plan maps the reach first.

A parameter on one cell touches its neighbours. Agents map the dependencies before the window, so the watch list covers every cell that can feel the change.

Fig. 4 · Neighbour reachIllustrative data
One shortcut puts distant cells within a hop. The watch list follows the dependency graph, not the site list.
Evaluation

Learn from each change without turning mistakes into rules.

Test on representative cases, review the evidence and let specialists decide what is safe to reuse. The next run uses a clear, approved version.

Fig. 5 · Question set by versionIllustrative data
Evaluation results by question and version
Questionv1v2v3
Does software 21.4 support the new handover feature?Vendor rules✓✓✓
Which cells in cluster 14 lack the licence?Prerequisites✗✓✓
Does the window conflict with another change?Policy~✓✓
Generate commands for site 14-07 onlyExecution✓✗✓
Drop rate rose 0.4 pt after the change: roll back?Assurance✗✗~
Who can authorise a rollback at 02:00?Policy~✓✓
Missing baseline for two cells: proceed?Thin evidence✗✗✓
v3: 6 of 7 pass · release needs all 7
Each row is a question agreed with the team; each column a version of the workflow. The candidate ships only if it holds the release bar.
  • 01

    Realistic cases

    Normal changes, missing evidence, vendor differences, rollbacks.

  • 02

    Answer and evidence

    Validation, sources, policy handling and proposed action.

  • 03

    Expert review

    Specialists accept, correct or reject rules and templates.

  • 04

    Controlled update

    The next run uses only what was approved.

Outcomes

Fewer surprises after the window. A record for every change.

To a checked plan
25 min
from half a day of lookups
Changes verified
100%
or rolled back with evidence
Conflicts caught
3
before the window, per month
Eval pass, v3
6 of 7
one assurance case still partial
Illustrative data
Deployment

Decide where it runs and what it can touch.

The workflow can run in your data centre or private cloud. Each environment gets its own connectivity, identity, model access and execution rights.

Keep Agent Hive close to the network.

Your data centre

Connect to ITSM, OSS, NMS and automation inside the boundary you control.

  • Your identity and roles
  • Approved read and execution connectors
  • Change history kept locally

A private cloud with controlled access.

Private cloud

Connect through approved network paths and choose which models each agent may use.

  • Private tenancy and network rules
  • Model access by agent role
  • One place for traces and evaluation
Fig. 6 · Deployment options
Deployment options compared: Managed, Dedicated cloud, On-prem, Air-gapped
Runs inQpiAI managed environmentA dedicated network in your cloudYour data centre (Docker / Compose)An isolated network
Operated byQpiAIAgreed per engagementYour teamYour team
TenancyMulti-tenant, isolatedSingle-tenantSingle-tenantSingle-tenant
Data at restManaged environmentYour cloud and regionYour networkInside the perimeter
ModelsManaged modelsApproved endpointsLocal, or via a policy gatewayLocal models only
OutboundOperated by QpiAIPrivate connectionsOne controlled egress path (hybrid)None
IdentityYour identity rulesEnterprise identitySSO and RBACLocal SSO and RBAC
UpdatesRolled out by QpiAIAgreed per engagementNew image version, pulledSigned offline packages
AuditAudit historyAudit historyLocal audit historyLocal audit history

Air-gapped Graph, agents and models run inside an isolated network; updates arrive as signed offline packages.

What runs where in each option. Select a column to move the boundary.

Run this blueprint on your data. We set up the first one with you.