Independent engineering. Connected operations.

Your systems.
One operation.

We connect billing, call control and AI so your business can work as one.

See how it connects

Keep what works.
Connect what doesn’t.

Your portal, switch and billing system can stay. We build the interfaces and workflows between them.

A customer should not need three teams
to activate one service.

01

From request to service

Connect the portal to provisioning, routing and billing. Make each handoff visible.

02

From call to explanation

Bring call records, quality evidence and charges into the same operational picture.

03

From manual work to useful AI

Give assistants approved data and tools, with people in control of consequential actions.

Understand the workflowBuild and prove itRoll out togetherSupport your team

An integration we are working toward

Follow one call.
See the whole business.

Hadlo’s portal. Sergey Popov’s billing and provisioning. Telegent’s SIP Control Plane. One connected workflow.

01

Prepare the route before the call.

The billing platform prepares the commercial policy. SCP validates it and publishes a version to the routing runtime.

Planned architecture
Planned Hadlo flow: portal changes pass through billing and SCP to local routing; call events feed billing and the operator console.
Business policy moves into the switch. Call evidence comes back to the people running the business.
  1. Portal requests a change.
  2. Billing prepares the business policy.
  3. SCP validates and publishes it.
  4. The local runtime serves routing decisions to OpenSIPS.
  5. Durable events feed billing and the operator console.
Try the call flow and a carrier timeout

Illustrative browser demo. Synthetic data, no live calls or carrier requests.

Customer & operator UI

Hadlo portal

Number search, assignment, routing and reports for each customer.

Existing · owner’s system

Contract & orchestration

Integration layer

Idempotent commands, tenant checks, one owner per identifier, explicit errors.

Proposed

Business operations

Billing · provisioning · LCR

Sergey Popov’s platform: rates, least-cost routing, provisioning, rated records.

Existing · owner’s system

Permitted AI

Operator assistant

Explains decisions from evidence; proposes changes; acts only through the contract and its role.

Prototype · simulated models

Signalling & admission

SIP control plane

Versioned routing, admission limits, edges. Full integration into this stack is planned.

Planned · from source-project experience

Numbers & trunks

Carrier adapters

Vendor APIs behind one interface. In the lab: fake vendors only.

Lab

Evidence

Durable events & CDR reconciliation

Replay-safe call legs; duplicates and late events handled; totals must match.

Proposed

Quality

Voice-quality evidence

Test-call measurements attached to the call they describe.

Lab
  1. Search & assign. An operator searches numbers for synthetic tenant “Demo Carrier A” and assigns one. The request carries an idempotency key.number.assign.requested · key=7f3c
  2. Reserve in provisioning. The number is reserved in the provisioning platform , its owner. A retry with the same key returns the same reservation.number.reserved · DID +1 555 0142 · tenant=demo-a
  3. Order from the vendor. The fake vendor confirms the order. No real vendor is contacted in the lab.order.confirmed · vendor=fake-1
  4. Publish routing. A new routing version with the current LCR rate deck is published; the previous version is kept so the agreed rollback procedure can restore it.routing.published · version=42 · rates=r7
  5. Admit a synthetic call. The call is admitted within its limits, routed by LCR, and carries the routing version with it.call.admitted · route=B · routing_version=42
  6. Record the call legs. Legs are written as durable events. A duplicate leg sent twice is ignored by its key.cdr.leg ×2 · duplicate ignored ×1
  7. Attach quality evidence. A synthetic test-call measurement is linked to the same call.vq.sample · synthetic · linked to call
  8. Rate and reconcile. The rated record and the switch legs are reconciled. Replaying the events adds no charge.reconcile.ok · legs=2 · charge delta=0 on replay
  9. Report back. The operator sees the number active, the call completed and the reconciliation status , one view, three systems.report.ready · tenant=demo-a
  10. Explain, don’t act. Asked “why route B?”, the assistant answers from the recorded events. It cannot change routing without an approved command.assistant.answer · sources=3 · actions=0
What is running, and what is planned?

The portal, billing platform and SIP Control Plane are separate source systems. The combined Hadlo workflow shown here is a proposed integration, not a completed deployment. The walkthrough is simulated. The brief builder runs in your browser.

A source system’s working feature does not establish that its integration is finished.

Built for one business.
Useful to the next.

Each integration should leave tested components, documented interfaces and a team that knows how to use them.

A client need becomes a proven workflow, then a tested adapter in the shared library, ready for a future integration.
  1. A client brings a need.
  2. The team proves a workflow.
  3. The team tests an adapter.
  4. The library keeps the permitted module.
  5. The next integration reuses it.

We build the connection.
Your team learns to run it.

Engineering, monitoring, staff training and support belong in the same engagement. Reuse is limited to components and knowledge we are permitted to carry forward.

Start with a workflow.
Grow from a working result.

What would you like to connect?

Choose the systems and the result you need. Download a project brief for the first conversation. Everything stays in your browser.

Good systems need people
who see the connections.

Engineers, operators, analysts and teachers: bring a concrete problem, a useful skill and work others can build on.