Ready to scale up? Find out before it breaks

A seed-stage startup needs product-market fit, not a platform team. Every hour spent on Kubernetes before customers pay is an hour not spent finding out what they will pay for. Platform engineering starts to pay back later, once the product sells and growth begins to strain what one or two engineers set up in a hurry.

That later stage is the scale-up. Customers now ask for uptime commitments and security questionnaires, the team has doubled, and the person who built the infrastructure is still the only one who understands it. The questions below help you see which parts of your setup are ready for that, which are not yet, and which have already waited too long.

Startup or scale-up: when DevOps starts to matter

The signs that you have crossed over are rarely technical at first:

  • Enterprise customers send security questionnaires and ask about your SLA before they sign.
  • An outage now costs revenue and reputation, not just an awkward afternoon.
  • Your engineers spend more time on infrastructure than on the product.
  • One engineer handles everything from deployments to night-time alerts.
  • Cloud costs grow faster than revenue, and nobody can say exactly why.

If three or more of these sound familiar, read the dimensions below with your own setup in mind.

Book a Consultation

The readiness framework

Each dimension lists the signals of being ready, of not being ready yet, and of having waited too long. Count where you land.

Reliability and uptime

  • Ready: you know your availability over the last quarter, and it meets what your contracts promise.
  • Not yet: nobody has measured availability, so the SLA in your contracts is a hope.
  • Waited too long: customers have started asking for credits after outages.

On-call and 24/7 coverage

  • Ready: the rotation survives holidays and illness, because at least four people share it, as Swiss on-call rules require for an unbroken rotation, or an operations partner covers the hours your team cannot. Common alerts have written runbooks.
  • Not yet: one person answers every alert, and holidays are planned around them.
  • Waited too long: an incident went unanswered overnight because that person was asleep or away.

Security and compliance

  • Ready: you can answer a customer's security questionnaire from documentation you already have.
  • Not yet: each questionnaire takes days of engineering time to answer.
  • Waited too long: a deal stalled on compliance requirements such as ISO 27001, GDPR or FINMA expectations.

Scalable architecture

  • Ready: environments are defined as code, and a new one comes up without manual steps.
  • Not yet: staging differs from production in ways nobody has written down.
  • Waited too long: a traffic peak took the service down because capacity could not grow in time.

Observability

  • Ready: monitoring tells you about an incident before a customer does.
  • Not yet: you have dashboards, but no alerts tied to what users experience.
  • Waited too long: customers regularly report outages to support first.

Cloud cost control

  • Ready: you know the cost per environment and per customer, and someone reviews it monthly.
  • Not yet: the cloud bill is one number that grows each month.
  • Waited too long: cloud spend has grown faster than revenue for several quarters.

Release velocity

  • Ready: every change goes through an automated pipeline, and a release is a routine event.
  • Not yet: releases need a specific person and a quiet evening.
  • Waited too long: the team batches changes into large, risky releases because each one is painful.

Team and knowledge

  • Ready: infrastructure decisions are documented, and more than one person can change production safely.
  • Not yet: the knowledge lives in one or two heads.
  • Waited too long: the person who built the platform has left, or has given notice.

AI readiness

  • Ready: AI features and agents go through the same pipeline, monitoring and access controls as the rest of the product.
  • Not yet: AI experiments run on a separate setup that nobody operates.
  • Waited too long: an AI feature reached customers without the release and monitoring discipline the rest of the product has.

AI in production rests on the same foundations as every other workload: automated releases, infrastructure as code and observability. The dimensions above are what make it possible.

Book your scale-up readiness check

The readiness check is a conversation with a VSHN architect about where your setup stands, at no cost. For a written assessment, the Application Deployment Review is a collaborative workshop with VSHN solution architects that assesses your current setup and recommends a target architecture, starting from CHF 1,900 for a single codebase. VSHN provides a scope and cost estimate within one business day.

Book your scale-up readiness check

Scale-up readiness FAQ

When does a growing software company need platform engineering?

When growth starts to strain the setup, not before. The usual triggers are customers asking for uptime commitments, a single engineer carrying all infrastructure and on-call work, and cloud costs growing faster than revenue. Before product-market fit, the time is better spent on the product.

What happens in the readiness check?

A VSHN architect goes through your goals, current infrastructure and constraints with you, using the dimensions on this page. The conversation is free. If it shows that a deeper look is worthwhile, the Application Deployment Review turns it into a written assessment with a recommended target architecture.

How much does an assessment cost?

The first conversation costs nothing. The Application Deployment Review starts from CHF 1,900 for a single codebase. Consulting beyond that is billed at CHF 250 per hour, and every engagement starts with a written scope and cost estimate.

We already have a DevOps engineer. Is this for us?

Often, yes. One engineer can build a good setup but cannot cover 24/7 on-call alone or take a holiday without leaving production uncovered. The check shows where that engineer's work holds up and where more people, or an operations partner, close the gap. An unbroken 24/7 rotation needs at least four people under Swiss on-call rules.

What if we are late in several dimensions?

Start with the one closest to customers: availability, on-call and observability usually come first, because those failures are the ones customers see. The readiness check helps you decide the order, and the Application Deployment Review gives you a target architecture to work towards.

Does VSHN take over operations after the check?

Only if you want it to. Many teams use the assessment to plan their own work. For ongoing operations, VSHN's Custom Application Operations Support (CAOS) is a retainer from CHF 800 per month per application. It covers documentation, regular reviews and readiness to support; manual effort is billed by the hour, and application monitoring and alert handling are not part of it. 24/7 incident response is the separately billed 24/7 OnCall Support plan, on which VSHN typically responds within one hour to critical incidents.

Talk to a DevOps expert

Ready to discuss your infrastructure, platform engineering, or operations needs? Book a consultation with a VSHN architect. No commitment required. You will receive a scope and cost estimate within one business day.

Book a free call

Or ask your question