Skip to content
luftflow

Keep the code, change one URL

Your short code, aggregator contract and APIs stay as they are. luftflow answers beside your service on real calls until the two agree, and going back is one paste.

A man in a patterned shirt talks on his phone.

What changes

One line of configuration at your aggregator.

Stays as it is

  • Your short codeThe same code, rental and agreement.
  • Your aggregatorUnchanged, and billed separately from luftflow.
  • Your APIsCalled by the flow at the same URLs, stubbed until you are ready.

Moves to luftflow

  • The callback URLPoints at luftflow, with separate addresses for staging and live.
  • SessionsHeld by the luftflow runtime, hosted or on your network, with retries answered from the record.

The move, a week at a time

About a month, with callers on your old service until the last step.

Week 1: rebuild
We rebuild your menu screen by screen, and you walk it on the phone in your browser.
Week 1: tests for the paths you fear
Every path becomes a scenario that must pass before publishing, the fragile ones first.
Week 2: staging
Staging gets its own callback address on your aggregator's test code, dialled from a real handset.
Weeks 2 and 3: answer beside the old service
Callers get the old reply while luftflow's is kept a week for comparison, and writes are refused unless named safe.
Week 3: rehearse the release
Publish, roll back and publish again, so release day is a repeat.
Week 4: switch
The live callback URL moves on a quiet morning.

Rehearse on real calls first

luftflow answers beside your live service, and you read where the two differ before anyone switches.

Callers never notice
Every caller gets your old service's reply, untouched.
Every screen compared
Both replies are kept for a week, with the ones that differ at the top.
Nothing is charged twice
Calls that would change anything are refused unless you name them safe.

Where you are coming from

From Africa's Talking or another aggregator
Africa's Talking, MTN and Zamtel: your test code stays on the old service while you build, and the live code moves last.
From your own callback
Your handler is the specification: we rebuild its screens, keep your endpoints, and you stop maintaining the rest.
From an older engine nobody wants to touch
The converter reproduces how the old engine actually ran your flows, and reports where that differs from the YAML.

Going back is one paste

Your old callback URL is still yours. Paste it back and the next caller is on your old service.

What we need from you

  • Whoever owns the callback URL today, on one call
  • Your menu in any form: YAML, screenshots or the code to dial
  • A test endpoint for your APIs, or permission to stub them
  • Your aggregator's test short code, if you have one

Bring us your menu

Tell us what you run. If it is live, we will rebuild a path and send you a link to dial.

Request early access