Cloud-native AI agents and enterprise ML automation for Seattle and Eastside teams. Built in your Azure or AWS estate. Remote-first from the EU, per project.
Seattle is a cloud town, and it shows in how its companies buy. With Amazon anchored in South Lake Union, Microsoft across the water in Redmond, and an ecosystem of enterprise software firms, consultancies and cloud-adjacent startups spread across Bellevue and the Eastside, the region has spent twenty years absorbing platform discipline. The Allen School at the University of Washington keeps a machine-learning talent pipeline running through it, and the older economy (Boeing and aerospace suppliers, the Port of Seattle and its logistics chain) supplies the heavy, data-rich operational problems that make that talent useful.
The consequence is a buyer who does not want a clever script. Seattle engineering organisations have platform standards, and they expect anything you build to respect them: infrastructure as code, deployment through the existing CI/CD pipeline, telemetry into the observability stack the on-call rotation already watches, identity through the corporate directory, and a cost story that finance can read. An AI automation that arrives as a black box outside all of that is not a solution. It is shadow IT with a friendlier name, and it will be switched off the first time nobody can explain what it did.
EIGHTY8 is a Netherlands-based, remote-first AI automation and consultancy studio that builds the way your platform team already works. We deploy inside your Azure or AWS environment, define the infrastructure in code, wire agents into your data platform rather than copying data out of it, and put governance on inference cost from the start instead of after the first surprising bill. We are candid about the distance: the Pacific morning is our late afternoon, so we run async-first with written handovers and a short daily live window. There is no Seattle office and no US phone number, and every engagement is quoted per project rather than per seat.
Yes, and we prefer it. We deploy into your cloud accounts, define the infrastructure in code, ship through the CI/CD pipeline your team already uses, and emit telemetry into your existing observability stack. Your data stays in your tenancy, identity runs through your corporate directory, and when the engagement ends your platform team owns something they can read, review and operate.
By treating cost as an engineering constraint rather than a monthly surprise. Route the easy majority of requests to a smaller model and reserve the expensive one for the cases that need it, cache aggressively, batch what can be batched, cap and alert per workflow, and measure quality against an evaluation set so a cheaper configuration can be adopted with evidence instead of hope.
That is the usual shape of the work here: agents and automations that read from the warehouse or lakehouse your data team already maintains, write back through the interfaces they control, and respect the lineage and access model that exists. We build alongside your platform, not a parallel copy of it. A second uncontrolled copy of the data is a liability, not an architecture.
Async-first, in writing. There is a short live window where your morning meets our late afternoon for reviews and decisions, and outside it you receive a written end-of-day handover covering progress, decisions and blockers. We are a Netherlands-based studio with no US office, and we would rather set that expectation now than pretend the Pacific time zone is a detail. Pricing is per project, based on scope and complexity.
EIGHTY8