About

A product studio that ships, then keeps the pager.

apbld is a product engineering studio. We build applications on any tech stack — AI and agentic systems, multi-tenant SaaS platforms, and startup products taken from an empty repo to paying customers. We work out of Dallas–Fort Worth, Texas, and most of what we know came from operating platforms at scale rather than reading about it.

77
Platforms shipped
7489+
Services in production
700+
Tenant databases operated
21k
Documents embedded for AI retrieval
How we think

Opinions earned the hard way.

We operate what we build

Our own products are live, billed and supported. Running software in production is a different education from delivering it and walking away — and it is the reason our architecture choices are conservative where it counts.

Depth before breadth

One module finished completely, including the boring long tail, before the next one opens. A product with four finished modules beats one with twelve half-built ones every time.

AI where it changes the answer

Vision models reading construction plans into priced quantities is AI worth building. A chat box that summarizes your own dashboard back to you is not. We are direct about which one you are asking for.

The unglamorous half is the job

Tenancy, billing, permissions, migrations, backups, deliverability. Nobody demos these and every product lives or dies by them.

How we got here

Three chapters, one throughline.

Everything we know about building products came from operating them, usually at a moment when something had gone wrong.

  1. The platform years

    Enterprise ERP at scale

    A multi-tenant ERP for construction and manufacturing grown to more than thirty backend services on Kubernetes, serving over a hundred per-tenant Postgres databases, with OCR pipelines feeding a vector store for in-app AI chat. It taught us what running infrastructure at scale actually costs.

  2. The rebuild

    Retailer OS and BuildAEC

    Rather than modernise thirty services in place, we rebuilt the capability AI-first as two focused products — and migrated the customers across with per-tenant reconciliation gating every cutover. Both are live, billed and supported today.

  3. Proving the range

    EzyEvite and client work

    A free consumer invitation product shipping in seven languages, alongside sites and systems for consultancies, trade contractors and community organizations. Different scale, same discipline.

Track record

From a 30-service enterprise platform to a consumer app that ships in seven languages.

Years of running enterprise infrastructure at scale is where this studio learned its trade. Retailer OS and BuildAEC prove we can rebuild that capability AI-first and sell it. EzyEvite proves we can do consumer. The client work proves we can do it at any size.