Network service automation and orchestration for carrier-grade MPLS / EVPN networks. AutoLinx Carrier takes service orders from BSS/OSS, designs the circuit, allocates resources, pushes configuration to multi-vendor devices, verifies the result and rolls back automatically on failure. Everything, including the AI, runs on-premise.
A single change request is translated into each vendor's own syntax, held for approval, then applied to every in-scope router at once with backup, verification and rollback built in.
Carrier-grade service automation for MPLS networks, managing multi-vendor equipment from one system.
Multiple vendors and models in one system; new models are added through templates.
Processes circuit orders in parallel, from receipt through verification.
Dedicated least-privilege accounts; every command is traceable. Telnet only for legacy devices without SSH.
Design endpoints, path and parameters, then provision from the same screen.
Extend automation in standard languages; templates, configs and API payloads in standard formats.
Compares stored configuration with live device state on a schedule and reports mismatches.
Provisions across IS-IS domains in a single order and keeps working throughout a phased migration.
Classifies devices by domain, IS-IS level and role (P, PE, ABR/Border) from AutoDiscover data, and alerts before provisioning if a role or domain is missing or mismatched.
End-to-end across Access, Aggregation and Core, with per-device templates chosen by role and domain and cross-domain verification.
Works on Flat L2 and Unified MPLS during migration, choosing the method from both endpoints' architecture; if none applies, reports why to BSS/OSS.
Updates migrated devices via re-discovery or import without losing existing circuits.
Receives service orders from BSS/OSS through a northbound API, carries them to completion, and reports status back automatically at every step.
Accepts orders from the order system and BSS/OSS over a REST API and reports status back automatically.
Creates new circuits from orders and modifies bandwidth/QoS, VLAN, or port and device assignment, flagging whether each change affects service.
Suspends and resumes service without removing the core configuration or releasing resources.
Removes configuration from every device on the path, returns VLAN, RD/RT, Service ID and IP to their pools, then verifies that no configuration remains and records the result as evidence.
Tracks every order through standard states, reports each transition to BSS/OSS, and keeps a full auditable history.
Cancels an order mid-execution and restores network state and resources correctly.
Per service type: All-or-Nothing rolls back the whole order; Hold-and-Retry retries only the failed part for a set count or period, then rolls back and alerts.
A repeated order never creates a duplicate circuit or allocates resources twice.
Registers each new circuit with the monitoring system automatically once it is built.
Every change runs Backup → Dry-run → Push → Verify, with automatic rollback. There are no exceptions.
Backs up the running configuration of every affected device before each push, tagged with timestamp, order ID, device name and requester.
Keeps configuration history and compares any two versions.
Builds a candidate configuration and shows its diff against the running config for review before commit.
Validates commands for the device type and software version before anything is sent.
Schedules execution within defined maintenance windows.
After each push, checks AC interface and VLAN state, pseudowire/EVPN service state (VPWS, VPLS, L3VPN) and endpoint pairing, BGP sessions, LDP/SR, and end-to-end reachability.
Records pass/fail with technical detail and reports it back to BSS/OSS.
Rolls back every device in the order when a push fails or verification does not pass.
If rollback itself fails, raises a critical alert showing the latest state, the stranded configuration and the manual remediation steps.
Rolls back to any stored version through an approval step.
Logs every step of a change: requester or source system, order ID, device, commands, diff, verification and rollback results, and per-step timestamps. Stored immutably and exportable.
Allocates and releases network resources automatically through a defined lifecycle, with no duplicate allocation even under heavy concurrent load.
Manages IPv4 and IPv6 for management, loopback and service/customer addressing.
Manages VLANs within a defined scope.
Manages Service Instance ID / VC ID / EVI to pair circuit endpoints, and RD/RT according to operator policy.
Administrators define ranges, allocation policy and reserved ranges within their permissions.
Reserved on order receipt, assigned on successful verification, released on terminate or rollback.
Releases expired reservations automatically and holds released resources in a configurable cooldown before reuse.
Allocates atomically, so concurrent orders never receive the same resource.
Checks AutoDiscover data for live usage before allocating, such as VLANs already in use on the target port.
Reconciles pool records against the live network on a schedule and issues a discrepancy report.
Imports existing resources and circuits in full before go-live.
Reports resource utilization and alerts when a threshold is crossed.
API to query, reserve and release resources, with per-caller access control.
Covers L2 and L3 services on current EVPN technology as well as existing LDP and VPLS services.
EVPN-VPWS (RFC 8214), single-homed and multi-homed, plus LDP-based VPWS (Martini) interoperating with existing services.
EVPN E-LAN, VPLS (LDP-based and BGP-based) and EVPN E-Tree.
MPLS L3VPN (MP-BGP) with VRF, RD/RT and customer routing by static and BGP, plus EVPN IRB for combined L2/L3 services.
Stitching between EVPN-VPWS and LDP pseudowires where devices support it, and EVPN multi-homing with ESI.
Multiple customer-side AC types, with end-to-end MTU consistency checks and an alert on any mismatch.
Per-circuit policing and shaping, and CoS marking set or preserved.
Administrators add new service types through the service catalog and templates without changing the core. Includes a service matrix by vendor, model and software version.
Built for regulated networks: no plaintext credentials anywhere, and every action is auditable.
Stores device credentials in a secret manager, encrypted at rest and in transit. No plaintext in the database, files, logs, screens or exports.
Rotates credentials without disrupting operations, with separate credentials per device group or domain.
Provisioning accounts are restricted at command level and work with the operator's existing AAA.
Four standard roles, with fine-grained permissions per menu and action.
Authenticates against AD/LDAP and single sign-on, with session timeout, forced logout and a limit on failed logins.
Records every action (login/logout, provision/rollback, approvals, changes to templates, workflows, catalog or pools, API calls) with actor, target, time, source IP and result. Append-only and forwarded to SIEM.
APIs over TLS 1.2+, authenticated by JWT/OAuth 2.0 or API keys with expiry and revocation, plus rate limiting, IP allowlists and per-caller permissions.
Versioned APIs, delivered with a complete API specification.
AI runs entirely on the operator's own infrastructure. It recommends and explains; every device action requires an engineer's approval.
Rule-based reasoning and AI-assisted recommendations assess configuration and parameter compliance against policy and network state.
Workflows with conditional branches and parallel steps.
Correlates events and indicates the probable cause.
A daily report summarizing network status and events.
All AI processing stays on site: no configuration, topology, customer data, model updates or telemetry leave for the cloud. Models are updated offline through an approved process with validation before use. Compute requirements are stated up front.
AI output is a recommendation only. Every device action goes through an approval workflow.
Shows the rules or data behind each recommendation and logs every decision (accept, reject, modify) in the audit trail.
AI can be switched off while core workflows and provisioning continue normally.
Our engineers in Bangkok handle installation, training and support for the life of the system.
Templates, the service catalog and introductory automation scripting.
Permissions, user accounts and passwords, and database access control.
Creating API keys, configuration, and integrating external systems.
Configuring and tuning the AI, and designing workflows and operational rules.
A service and support organization in Thailand, providing guidance during cut-over and normal operation.
Notifies customers of every patch and new release or version, with release notes.
Installation, operation and maintenance manuals in Thai or English, plus solution architecture documentation.
Every order backed up, verified and reversible. Talk to us about a pilot on your MPLS / EVPN network.