Services: Build new, or modernize what you have
Your business runs on rules and processes that took years to get right. We codify them into one system built around how you work. Sometimes we build it new. Sometimes we modernize a legacy system without a rewrite, onto a stack built for the next decade. Either way it’s specified, tested, and documented before we hand it over.
Custom builds: Your domain, built as one system
We learn how your business works, write it down as specifications, and build production-grade software around it, with clean repos, modern stacks, and no lock-in. From day one, you control the roadmap and how it evolves.
How our builds differ
- Specified first
- Before any code, we write your domain down as specifications: access control, data model, business rules, and a journey for every role. You approve what gets built, and you keep the specs.
- A head start from our Foundations
- Login, permissions, reporting, and more are already built and hardened. The budget goes to the rules and workflows that make you different.
- AI-ready from day one
- LLM features run inside your app, behind your permissions and grounded in your records. The project ships with the specs, rules, agent roles, and tests your team’s AI tools need to extend it safely.
- Options on how we engage
- Usually a fixed-scope build. In some cases, for the right greenfield product, a build and revenue share, where ownership and terms are set in the agreement.
- You own the code. Code, CI, and docs live in your organization’s repo from the first commit. No vendor lock-in.
- One-time build, no licenses. No per-seat fees, annual renewals, or forced upgrades on the software we build for you.
- Self-host first. Run it in your own cloud account, which we can operate for you, or on-premises with your own IT staff.
- Built for developers and AI. Rust, OCaml, and Elixir, with specs, tests, and rules your AI tools can follow.
- Support on request. Support and new features, scoped and priced when you need them. No retainer.
- Transparent and extensible by design. Clear documentation, modular architecture, and no black boxes, so your team can extend it as your needs grow.
Modernization: Your system, modernized. Still yours.
Your legacy code holds years of business logic your teams earned: the rules, the edge cases, the fixes. A rewrite throws that IP away. Modernizing keeps it. We move the system to Rust, OCaml, or Elixir and keep its behavior, data, and contracts. Parity is proven with your own tests, compared permissions, and byte-compared outputs, and cutover goes module by module with a route back.
Included in delivery · foundations: The groundwork, already built
Our eight foundations, from login and permissions to AI middleware, come with every build and modernization, with full source under our license. They’re also the fastest way to modernize part of an existing system without modernizing all of it.
AI engineering: Faster with AI, safer by design
We use AI to move faster, with senior engineers reviewing everything, and we pick stacks whose compilers reject whole classes of bugs. Security comes first, because attackers have the same tools.
Optional module · Beacon: Competitive intelligence, wired in
Beacon is our competitive intelligence module. Wired into a system we build, it supports decisions and gives early warning: it reads what the world publishes about your market, answers the questions you define, and feeds alerts and data into the workflows that need them. It’s also available on its own.
Vendor apps: Own the system instead of renting one
We also replace vendor apps that don’t fit with a system of your own, built around how your team actually works.
How we engage: Three ways to work with us
Fixed-scope build
New software, specified up front and handed over complete.
Modernization
Your legacy system moved to a modern stack, module by module, with a fixed scope per module.
Build and revenue share
In some cases, for the right greenfield product, we build in exchange for a share of revenue instead of a straight fee. It’s selective, and ownership and terms are set in each agreement.
Security and your data: Your accounts, your data, on infrastructure you already trust
We don’t build our own data centers or security perimeter. We design each system to use the security, provisioning, and best practices of the largest cloud providers. AWS, Microsoft Azure, and Google Cloud invest more in physical and network security than any single firm can, and we build on them rather than around them. Your data and your IP get the same care when a system talks to AI models.
- Your accounts, your data. Your system runs in cloud accounts your organization owns. The data stays there, and you decide who has access, including us.
- The provider’s security, used properly. Physical and network security, access control, backups, redundant power, and compute failover come from the provider, set up to its own best practices rather than a home-grown version.
- Any major cloud, or on-premises. We are not tied to one vendor. We build for AWS, Microsoft Azure, or Google Cloud, whichever your organization already trusts. Or we deliver a system that runs on-premises, handed to your own IT staff.
- We can operate the cloud installation. We provision the infrastructure in your accounts and can act as your DevOps team from launch, under your access rules. The accounts stay yours, and we never host a client’s system on infrastructure of our own. The one exception is Beacon’s data service, which we operate. On-premises systems are run by your IT staff.
- Confidential details are removed from AI prompts. Names, account numbers, and other fields you mark as confidential are removed before a request leaves your system, and restored in the reply. The AI middleware that does this is part of the foundations we license.
- The right model for each request. The AI middleware’s router sends each request to a private or a public model based on how sensitive it is, what it costs, and how quickly it needs an answer. The most confidential work can stay on a private model, and every call is recorded, so you can see what each AI workflow sends, to which model, and at what cost.
Not sure which fits?
Tell us what you have and what’s in the way. We’ll recommend a new build, a modernization, or leaving it alone.