Say the number
If we do not know, we say we do not know. If a target is aspirational rather than measured, it is labelled as such. Marketing numbers that fall apart in a technical review cost more than they win.
The premise
Delivery and defence are one problem wearing two invoices
The industry split them for commercial reasons, not technical ones. A CDN sells throughput and cache efficiency. A security vendor sells block rates and threat feeds. Both are measured on things that are easy to put in a quarterly report, and neither is measured on whether your worst request was fast and safe.
That split is invisible while everything works. It becomes the whole story during an incident, when the correlation between a routing change and a spike in blocked requests exists only in the head of whoever happens to be looking at both dashboards.
We took the other position. One control plane, one dataset, one team, one contract. It is a harder company to build and a much easier company to be a customer of.
WHAT THIS MEANS IN PRACTICE
- The analyst who detects a problem can change the edge configuration that fixes it
- There is no first-line queue whose job is to establish that you have a real problem
- One SLA covers the path, so no incident can become a jurisdictional dispute
- Delivery and security telemetry sit in the same store, correlated by default
Operating principles
Six commitments we will be held to
These are not values-poster material. Each one costs us something, which is how you can tell we mean it.
Your data leaves with you
Logs are exportable in full. Custom rules written for your estate are yours. We do not want to win renewals because leaving is painful.
Name the person
Every incident has an accountable analyst with a name, not a ticket number. Every rule change is attributed. Anonymity is comfortable for the vendor and useless for the customer.
Keep the second network
We will advise you to retain a provider we do not control, even though it lowers our share of your spend. Single-sourcing your delivery path is bad architecture regardless of who the single source is.
Help during the incident, sell afterwards
Emergency mitigation is not conditional on signing. You get the stabilisation and the written report either way. Some of those calls never become customers, and that is an acceptable outcome.
Write the post-mortem you would want
When we cause an outage we publish what happened, what we got wrong and what changed, in enough detail to be useful and without the passive voice.
Engagement model
How an engagement actually runs
Read the traffic
A week of logs or a traffic summary, and a session with your engineers. We come back with the profile, the weak points and a price. No cost, no sequence.
Run it in parallel
Configuration built and validated alongside your live path. You compare real numbers before anything moves.
Cut over in stages
Percentage-based traffic migration with halt conditions agreed in advance and a rollback that has been rehearsed, not just documented.
Operate and review
Continuous tuning, a named analyst on Command, and a quarterly review that covers what went wrong as well as what improved.
Company facts
The register entries
BRAND NAMESeguroCloud · A brand of Hoki Digital Pte. Ltd.
HEADQUARTERSSingapore
OPERATIONSFollow-the-sun security operations across three regions
FOUNDED2021
Careers
Who we hire
People who have been on the wrong end of an incident and remember it precisely. Delivery engineers who can read a packet capture, analysts who can write a rule and ship it, and support engineers who have never been asked to follow a script.
Send us something you have builtTest the claims before you believe them.
Every principle on this page is checkable in a technical session. Bring the hardest question you have about your current setup.