The platform division.
Helix AI Cloud exists to take the parts of Helix Cloud Solutions’ work that belong in a product rather than a project, and maintain them like products: versioned, documented, contract-bound, and honest about their own defects.
A division, not a separate company.
Helix Cloud Solutions is a Kenyan technology company working across cloud, data engineering, AI, cybersecurity and software development — mostly as engagements, for enterprises, financial institutions and public bodies.
Engagement work and product work fail in different ways. An engagement ends; a platform has to still be correct in three years, on somebody else’s infrastructure, after the people who wrote it have moved on. Helix AI Cloud is where the second kind lives, so it can be held to the second standard.
Six things a platform of ours has to have.
Not aspirations. These are the things we check before calling something a platform rather than a deliverable.
A version number that means something
Wire protocols are versioned separately from products, and adjacent releases interoperate during an update window — because a fleet does not upgrade all at once and pretending otherwise breaks somebody’s evening.
A published contract at every boundary
Where another team’s system integrates with ours, the shape of the exchange is a document they can generate types from, digest-pinned so neither side can change it silently.
A defect register, in public where it matters
Every platform page on this site carries a section listing what the product does not do yet. An operator who finds a limitation themselves stops believing everything else they were told.
Verification by driving, not by reading
Most of the real defects in this work were invisible to inspection and obvious on execution — a privilege escalation, a query the database would not parse, an access-control layer that had never once run. Text review passed all of them.
Guards that a test can tell apart from their own absence
A protection nothing can distinguish from not having it gets deleted by the next person tidying up, correctly and with a green test suite. So we pin the guard structurally, not just behaviourally.
Data boundaries enforced by code
Where a subscriber’s data may cross to their provider, the permission is a live condition inside the query, not a flag checked once and cached. Revocation should be a fact, not a promise.
Built in Nairobi, for operators here first.
The assumptions differ. Subscribers pay by mobile money. Many connections sit behind carrier-grade NAT, which changes what an ISP can even do for a customer. Support conversations happen on WhatsApp before they happen in a ticket queue. A platform designed for a market where none of that is true arrives needing modification before it can be used at all.
None of that makes the software regional. It makes the defaults right for the people we build with, and correct everywhere else.
Twenty years on the other side of this conversation.
Helix Internet Monitor exists to help a subscriber prove their provider is underdelivering. It was built by someone who spent two decades being the provider — in the NOC that received those complaints, and in the engineering teams that had to decide which ones were real.
2005 — a field technician, then a NOC
The career starts at an ISP selling connectivity to small and medium organisations: field work and cabling first, then customer support, then the escalation desk. This is where you learn that most reported faults are not where the customer thinks they are, and that almost none of them arrive with evidence.
2007 — satellite, and hubs on three continents
Network and systems administration for an upstream provider selling satellite internet and WAN services to ISPs across Africa and the Middle East, working with RF engineering teams to keep levels correct across hubs in Nairobi, London and Washington. A satellite link teaches the thing this product is built around: latency has a physical floor, and grading a line against an absolute scale tells its owner nothing.
2009 — core IP at a national carrier
Team leader for IP Solutions and Server Services at Kenya Data Networks, then the region’s leading public data network operator and now part of Liquid Telecom. Designing and running the core IP and server infrastructure, supporting the engineers who handled escalations from the NOC.
The work included technical lead on the first Google Global Cache node in East and Central Africa, hosted at KDN and distributing to other ISPs through the Kenyan internet exchange; planning the IP infrastructure to take up submarine fibre capacity as it landed; and overseeing the transition of a large share of the country’s institutions from satellite to fibre. Peering was arranged through the Johannesburg and London exchanges to cut transit cost and, more to the point here, to cut latency and hop count.
That is the same measurement this product now puts in a subscriber’s hands: how many hops, whose network each belongs to, and where the delay is actually added.
2012 — projects across Africa and the Middle East
Assistant projects manager at a satellite services operator, designing and supervising deployments for ISPs, enterprises and international organisations — including ground monitoring for a satellite fleet across Kenya and Tanzania, border-control network infrastructure for a UN agency, and the logistics and technical coordination of a BBC World Service programme carried by 73 partner stations. One earth station was brought back into specification after weeks of tuning, ending transponder charges the operator had been paying for a signal that was out of tolerance.
2019 — Helix Cloud Solutions
Founded to do this work as engagements: core and RF network overhauls for ISPs, security hardening, virtualisation, IP and ASN allocation, and the monitoring and ticketing systems an operator needs to run a network rather than merely own one.
Building those systems repeatedly, for operators who could see their own network but never their customer’s line, is where both platforms came from — and why this one is built on the subscriber’s side of the demarcation point.
Built by Helix Studios.
Helix Cloud Solutions has two arms that meet at a product. Helix Studios is the software development arm: it builds the platforms. Helix AI Cloud is the platform division: it runs them, sells them and answers for them.
The separation is not organisational decoration. A team that builds and then leaves is a team that never finds out which of its decisions were wrong; the six standards above exist because the same people operate what they wrote. Helix Internet Monitor is in the Studios portfolio and runs under Helix AI Cloud, and every limitation either of them knows about is published on this site rather than discovered by a customer.
Work with the division.
Whether you operate a network, sell connectivity, or just want to know what your own line is doing.