The product gets the attention. The admin tooling decides whether support scales.

Who this is for

Software companies building a product, and companies whose product is software even if they do not describe themselves that way. Early stage teams needing a first platform built properly, and established teams who need capacity for the work their own roadmap keeps deferring.

A deliberate limit on the claim: Goldstone builds custom products and platforms. That is not the same as being a venture scale SaaS specialist, and we are not presenting it as one.

What usually goes wrong

• Tenancy decided by accident. How organisations are isolated gets chosen in the first fortnight and is expensive to revisit once there are customers.
• Onboarding by hand. Every new customer needs someone technical, which caps growth at the attention of that person.
• Billing edge cases. Taking a card is easy. Plans, proration, dunning, refunds and the reporting that reconciles them is where the real work is.
• No support tooling. Support cannot reproduce a customer’s view, so every ticket becomes an engineering interruption.
• Deferred integrations. The thing customers keep asking for stays below the line because the roadmap is features.

The pattern is that product work is visible and operational work is not, so operational work loses every prioritisation meeting until it becomes urgent.

What Goldstone may build

• The product or platform itself, where that is the project
• Multi-tenant architecture with isolation enforced in the database rather than in application code
• Customer onboarding that does not need an engineer
• Billing and subscription workflows, including the unglamorous cases
• Customer portals and account management
• Internal admin and support tooling, including scoped impersonation with an audit trail
• Integrations that have been on the roadmap too long
• AI features inside the product, where they earn their running cost

On multi-tenancy specifically

Worth being concrete, because it is the decision most expensive to get wrong. Where it fits the system, Goldstone prefers tenant isolation enforced in the database through row level controls, rather than relying on every application query to remember the tenant boundary. It reduces the number of places a mistake can expose another customer’s data, and we have built it that way on client work.

If you already have customers on an architecture where isolation is application level, changing it is a real project. It is also one of the few pieces of technical debt genuinely worth paying down early.

AI opportunities

Inside the product, where it is a feature customers are paying for. Around it: support triage, documentation retrieval, classifying inbound requests. The same rule as everywhere, with the added consideration that per request cost becomes per customer cost, and that has to work at your pricing.

Where ordinary automation is better

Provisioning, entitlement checks, billing rules, usage limits and every scheduled job. These are rules with money attached, and rules with money attached should be deterministic and auditable.

When custom software makes sense

Almost by definition here, with one caveat: do not rebuild commodity infrastructure. Authentication, payments, email delivery, analytics and error tracking are solved. Build the product and the parts that are genuinely yours, and integrate the rest.

Typical engagement and investment

Product and platform work sits in the Custom Apps bands, with multi-tenant and software product builds scoped individually. Internal tooling is often a smaller business systems project alongside it.

Next step

Describe what you are building for a preliminary plan, or a Systems Blueprint if the architecture is the question rather than the build.