Kiapoynt
Nairobi · Software engineering

We build the systems
Kenyan businesses run on.
Then we run them.

ISP operations, mobile money, commerce and identity infrastructure. Software that talks to the router, reconciles the payment and opens the ticket — not another place to type up what already happened somewhere else.

See the work
Production · liveOperational
ISP tenant · instance 01
Optical polling · 30 min interval
Operational
ISP tenant · instance 02
Isolated credentials · circuit breaker
Operational
ONUs per poll cycle
Alerting + auto ticket creation
417
Ticket escalation
Time-based · reassigns support group
4 tiers
M-Pesa channels
Live in production
Paybill + Till
Figures reflect systems currently in production
Integrates withM-Pesa DarajaMikroTikFreeRADIUSSmartOLTSplynxNCBAAfrica's TalkingAWSGTM / GA4
The gap

The outage gets logged correctly. Nobody acts on it until morning.

This isn't a monitoring failure. SmartOLT sees the ONU go dark at the exact minute it happens — the record is right there. What's missing is the step after: something that reads that signal at 2am and acts on it, instead of waiting for a human to notice in the morning.

Below is the same real incident, once without that layer and once with it. The equipment and the failure are identical — only the four hours in between change.

A typical outageUnmonitored
  1. 02:14

    ONU on Node 14 loses signal. SmartOLT logs it correctly.

  2. 02:14

    Nobody is watching the dashboard. No ticket, no alert sent anywhere.

  3. 06:40

    A customer messages asking why they've been offline all night.

  4. 06:52

    An agent checks SmartOLT by hand and finds the outage.

  5. 07:10

    A ticket is opened manually in Splynx, referencing nothing.

4h 56mtime the customer was offline
Same outage, with KiapoyntAutomated
  1. 02:14

    ONU on Node 14 loses signal. SmartOLT logs it correctly.

  2. 02:14

    The next poll cycle catches it and opens a Splynx ticket automatically.

  3. 02:14

    Deduplicated against Node 14's history. Assigned tier 1.

  4. 02:44

    No update in 30 minutes — escalates to tier 2, wider group notified.

  5. 03:08

    Resolved.

54mtime the customer was offline

Illustrative timeline based on the ISP operations platform's actual poll interval and escalation rules — exact timings vary by network. See the full case study.

Systems we integrate with

Not a partner list. These are the APIs, protocols and hardware our production systems talk to every day.

M-Pesa Daraja
MikroTik
FreeRADIUS
SmartOLT
Splynx
NCBA
Africa's Talking
AWS
GTM / GA4
Proof

Numbers that belong to a specific system.

Nothing here is a round marketing figure. Each one comes from a system we can point at.

0

ISP operators running on the same platform, with isolated credentials and independent failure handling

ISP operations platform
0

optical network units polled per cycle, with alerting and automatic ticket creation

ISP operations platform
0

escalation tiers that reassign the support group and raise priority without a human deciding to

ISP operations platform
0

M-Pesa channels live in production — paybill and till — behind one checkout

Blessed Collections
How we run it

Shipping it is the easy half.

The systems we build handle money and network infrastructure, so the deployment path matters as much as the code. Nothing reaches production without passing the same gates.

How we work
  1. 01Docker Compose on AWS EC2
  2. 02Images to AWS ECR
  3. 03GitLab CI/CD
  4. 04Lint + tsc --noEmit gates
  5. 05Trivy vulnerability scanning
  6. 06Transaction-wrapped migrations
  7. 07Pre-flight database backups
  8. 08Per-tenant credential isolation
Questions

The ones people actually ask.

Can you integrate M-Pesa into a system we already have?

Yes. We work against the Safaricom Daraja API and have paybill, till and STK Push flows running in production, including bank webhook reconciliation. We can add payments to an existing system without rebuilding it.

Do you work with our existing billing platform?

Usually. We build on top of platforms operators already run — Splynx for ISP billing, SmartOLT for optical monitoring — rather than asking you to migrate off them. The integration is the product.

Do you maintain systems after launch?

Yes, and it is most of what we do. Our systems run on infrastructure we deploy and monitor: containerised on AWS, deployed through CI with vulnerability scanning, with backed-up and transaction-wrapped database changes.

Do you build ecommerce stores?

We build stores that take mobile money and report where each sale came from. Analytics and conversion tracking are wired in at build time rather than added later, so the store is measurable from day one.

Where are you based?

Nairobi, Kenya. We build for Kenyan and East African operating conditions — mobile money, local billing platforms, and hardware that is often several hours away from the nearest engineer.

Next step

Tell us what breaks, and we'll tell you whether we're the right people to fix it.

A short conversation about what you run today, where the manual work sits, and what it would take to close the gap. If it is not a fit, we will say so.

business@kiapoynt.com