What does Nrth AI build?

Nrth AI builds the Nrth Agentic OS — a secure, on-premise AI runtime that connects enterprise workflows, AI models, agents, and skills into one accountable layer — and ships paid early-access deployments.

How does engagement start?

We follow a three-step process: Map (1 week of discovery), Pilot (2–3 weeks of a working agent in production against a bounded slice), and Scale (ongoing hardening, observability, and expansion).

How much does it cost?

Engagements range from AI Employees (setup from ~$5,000) to on-premise infrastructure ($10,000–$100,000+) and custom development (from ~$15,000). Pricing and timelines are listed on the services page.

How do I get in touch?

Send a discovery brief at https://nrth.no/#contact, or email contact@nrth.no. A human from the Nrth AI team responds within 1 business day.

Where is Nrth AI located?

Bergen, Norway. We work with clients across the Nordics and beyond.

AI & AI Agents for enterprises · est. 2026

Your next hire isn't a person.

✦ Built on the Nrth Agentic OS.

Nrth AI builds custom AI & AI agents that handle the work your team shouldn't be doing. Production-grade, observed, and accountable — like the best employee you've ever had.

Send us a brief → Explore the platform ↓

Customer support agents ✦ Sales-outreach agents ✦ Research & intel agents ✦ Back-office automation ✦ Internal knowledge agents ✦ Voice & phone agents ✦ Customer support agents ✦ Sales-outreach agents ✦ Research & intel agents ✦ Back-office automation ✦ Internal knowledge agents ✦ Voice & phone agents ✦ Customer support agents ✦ Sales-outreach agents ✦ Research & intel agents ✦ Back-office automation ✦ Internal knowledge agents ✦ Voice & phone agents ✦

✦ 01 · Product

The Nrth Agentic OS.

A secure runtime that connects your enterprise workflows, AI models, agents, and skills into one accountable layer. Not a chat box bolted onto your data — a system that runs the work, inside your environment, under your control.

✦ How it works

Setup, workflows, agents, skills.

Layer 01Setup

Connect your systems, data, and models. The OS deploys into your environment — on your own hardware or your cloud — and maps the tools, permissions, and guardrails each agent is allowed to use. Nothing runs without an explicit boundary.

Layer 02Workflows

Define the work. Real business processes become structured, observable workflows the OS can run, monitor, and improve — not brittle scripts that break when an edge case appears.

Layer 03Agents

Put agents to work. Named agents execute the workflows end to end — calling models and skills, checking their own work, and escalating to a human the moment a task needs one.

Layer 04Skills

Extend what agents can do. Reusable skills — search, retrieval, compute, integrations — are shared across every agent and workflow, so capability compounds instead of being rebuilt each time.

✦ Why it's different

Secure by architecture, not by promise.

01Hardware

Runs on any hardware

A GPU server, a workstation, or the infrastructure you already own. The OS is hardware-agnostic — no lock-in to a single vendor or cloud.

On-prem or cloud

02Sovereignty

Data never leaves

Deploy on-premise and your data, prompts, and outputs stay inside your network. Nothing is sent to external servers you don't control.

Your walls · your data

03Security

Accountable by design

Permissioned tool access, full audit logs, evals, and guardrails on every agent action — with a human in the loop where it matters.

Audited · permissioned

04Ownership

You own the stack

Keep the model, the data, and the IP. Agents run as accountable members of your team — not a black box you rent by the seat.

Model · data · IP

✦ Roadmap

Honest about where it is.

Now · in development

Core runtime + pilot deployments

The runtime, workflow, and agent layers are running against real work on partner hardware — hardening where it actually has to.

Shared skills library + observability console

A growing library of reusable skills and a console for evals, audit logs, and guardrails — so teams can see and govern exactly what every agent does.

Later

Self-serve deployment

Packaging the OS so teams can stand it up in their own environment with shorter onboarding — beyond hands-on partner deployments.

✦ Early access

Get the OS into production today.

The OS is in active development. Pilot customers get early access through custom deployments — we put it to work against a bounded slice of your operation, with measurable outcomes, while it matures. Those engagements are how the first partners run it in production before general availability.

See the engagements → Send a brief ↓

✦ 02 · Work

Outcomes, not slideware.

Concrete numbers from agents running on the OS — what each one is built to hit against a bounded slice of real work. Proof the OS works in production, not a consulting portfolio.

72%

auto-resolved

5×

SDR throughput

FTE added

✦ The agents

Built for the work, not the demo.

01Support

Tier-1 support, handled

Agents that read your docs, ticket history, and CRM — resolve routine tickets in seconds and escalate the nuanced ones with full context.

72% auto-resolved

02Revenue

Inbound qualified, outbound personalized

Agents that research prospects, draft first-touch messages in your voice, and keep the pipeline warm while your team sleeps.

5× SDR throughput

03Ops

The invisible back-office

Invoice matching, vendor comms, compliance checks, data reconciliation — the boring, expensive work becomes a background process.

0 FTE added headcount

✦ How we prove it

Map. Pilot. Scale.

i

Step 1 — Map

One week. We sit with your team, shadow the work, and find the 3–5 places an agent will change the unit economics.
ii

Step 2 — Pilot

Two to three weeks. A working agent in production against a bounded slice — real data, real users, measurable outcomes.
iii

Step 3 — Scale

We harden, observe, and expand. Agents operate as named team members, with evals, guardrails, and a human in the loop where it matters.

✦ 03 · Company

Infrastructure for enterprise AI agents.

Nrth AI is a pre-seed startup based in Bergen, Norway (est. 2026), building the next-generation operating system for enterprise AI agents.

✦ Why now

The timing isn't an accident.

01Regulation

Regulation is forcing the issue

Regulated industries, manufacturing, energy, and maritime can't ship data to servers abroad — yet they're now expected to adopt AI anyway. That tension is the market we're built for.

Regulated · ready

02Economics

The "AI employee" math

A skilled employee in Norway costs 1–1.5 MNOK a year. An equivalent AI system costs 15–35% of that, works around the clock, and never churns.

15–35% of a hire

03Gap

Most teams get tools, not systems

The market hands people parts and a subscription. Very few deliver the finished, accountable system that actually does the work — which is exactly what the OS is.

Finished, not parts

04Reach

Borderless from day one

The OS deploys anywhere — cloud or a customer's own hardware. The need is real and global, so we're built to ship across borders from the start.

Deploy anywhere

✦ Team

Finance and delivery depth meets builder energy.

Thomas Uthaug

Co-founder · Business, product & delivery

A decade across finance, quality leadership, and technology delivery — Storebrand and Sopra Steria. MSc in Finance from NHH, with exchange studies at UC Berkeley and Copenhagen Business School. A career built on translating complex technology into bottom-line business value.

NHH MSc FinanceUC BerkeleyStorebrandSopra Steria

Simen Uthaug

Co-founder · Technical & developer

BSc in Information Science (UiB) plus a master's degree in UX and Development. Full-stack developer experienced in TypeScript, React, Python, API integration, databases, and machine learning (neural nets, NLP).

UiB Info. ScienceTypeScript / ReactPython / MLFull-stack

✦ 04 · Contact

Don't email us.
Send a brief.

Tell us what you want automated. A human on our team replies with a tailored proposal — usually within one business day.

Prefer plain email? contact@nrth.no

Bergen · Norway

nrth.brief — fill in

brief.v1

Brief sent.
We'll be in touch.

A human on the Nrth AI team replies within one business day from contact@nrth.no.