Services

Three ways we engage.

We are an operator-led practice: we run our own production systems, a food marketplace, a payments module, a property-data platform, and a case-management system for legal work. Every engagement applies the same agentic engineering method those products are built on. Retained engagements only, never hourly, never ad-hoc, a small number per quarter to keep the work serious. Pricing is scoped per engagement after a short discovery conversation. No rate card, no surprises.

AI readiness assessment

2–3 weeks · scoped per engagement

A grounded answer to where AI agents pay off in your operation, and where they create risk, before you commit to a build. Fixed scope, fixed price agreed before we start.

Deliverables
  • ▸Workflow map: which processes suit agentic automation, which do not, and why
  • ▸Data + compliance review: GDPR posture, confidentiality constraints, EU hosting
  • ▸Prioritised automation plan with effort and impact estimates per opportunity
  • ▸Exposure review against our rule library of 75+ codified production failure modes
  • ▸90-minute walkthrough with your leadership or engineering team, plus a written report you own
Best fit

Organisations at the start of AI adoption, anywhere remote. Danish companies can often have part of the cost covered through the SMV:Digital grant programme; we help with the application.

When the system prevents a single failure mode from shipping to production, it typically pays back the cost of the engagement that codified it.

Agentic build

8–16 weeks · scoped per engagement

We come in, ship one production system end-to-end with the agentic method, and hand your team the patterns to extend it.

Deliverables
  • ▸One production system shipped with the full agentic engineering method applied
  • ▸Project-scoped rule library, code-enforced where possible
  • ▸Audit-trail infrastructure and human-review-loop UI integrated into your stack
  • ▸MCP server scaffold or equivalent agent-native API surface
  • ▸Internal team enablement: pairing sessions, code review, methodology handoff
Best fit

Teams that need a wedge agentic system shipped reliably, with the patterns to do the next ones internally.

When the system prevents a single failure mode from shipping to production, it typically pays back the cost of the engagement that codified it.

Transformation programme

6+ months · scoped per engagement

Multi-system rebuild + internal team uplift + methodology transfer. The agentic delivery method becomes how your engineering org ships.

Deliverables
  • ▸Rebuild of 2–4 production systems with the agentic method
  • ▸Methodology embedded in your engineering rituals (review, deploy, post-mortem, planning)
  • ▸Internal champions identified, paired with us for the duration
  • ▸Long-term rule library and observability, owned by your team at handoff
  • ▸Quarterly outside-eyes reviews for 12 months post-handoff
Best fit

Series B+ companies, established SMBs in technical rebuild, public-sector tech transformations.

When the system prevents a single failure mode from shipping to production, it typically pays back the cost of the engagement that codified it.

What we run ourselves

The method is not theory. These run on it.

PickNDeal

B2B + D2C food marketplace. 14 phases shipped end to end: ordering, payments, AI agents, mobile apps.

PayoutKit

Stripe Connect module, extracted agentically from PickNDeal as a reusable payments component.

Boligrede

Decision support for Danish home buyers. Property registers, energy labels and deed data pulled into a structured buy decision, with the legal withdrawal window handled correctly.

Case management for legal work

A platform for document-heavy legal workflows, in development in collaboration with a Danish law practice. Under confidentiality.

How we work

Five things baked into every engagement.

Faceless team voice

We operate as a team, not a personal brand. Engagements are with the firm, not with an individual.

We meet your stack where it is

The methodology is stack-agnostic. We've shipped production systems in Python, JavaScript/TypeScript, .NET, Java, Node, PHP and most modern combinations. Our own products run on Next.js + Postgres + Drizzle + MCP-native + Anthropic SDK; your engagement runs on your stack. Your team won't inherit something they can't maintain.

Audit-trail-first

Every agentic action we ship goes through the audit-trail pattern. When a failure mode lands, the trail is how we diagnose without re-running the agent. When your CISO or board asks “why did it do that,” the trail is the evidence.

Methodology stays with you

The rule library, audit infra, and review-loop UI all live in your repo at handoff. You own what we built.

Confidentiality by default

Engagements run under standard NDA. Your codebase, data, business logic, vendor relationships and any details that would identify your company stay confidential. The only material that ever flows to our public output (journal, OSS, invariant library) is generalised technical patterns or anonymised lessons. Public case studies require explicit written consent from your side; the ones we currently publish are our own products (PickNDeal, PayoutKit, Boligrede), not client work.

Ready to talk?

Send a paragraph or two about the system you want shipped, audited, or transformed. We respond within two business days.

Start a discovery conversation