Portrait of Håkon Berntsen, Senior AI Architect
Available · Q3 2026

// SENIOR AI ARCHITECT · CTO · CEO

Production AI systems for healthcare, regulated industries and the enterprise.

I architect and operate the AI products behind several companies — all sharing the same cloud-native, GDPR-grade backbone. Today: CTO at Eir Tec LTD, CEO & Co-founder of Eir Tec AS, COO & Systems Architect at MediVox, and CTO at Skjld Labs.

See the work Book a call

// Running in production

  • MEDIVOX
  • EIR TEC LTD
  • EIR TEC AS
  • SKJLD LABS
  • JAYDUS.AI
  • INFODESK
  • OPEN INFO

// 02 — Portfolio

Selected Work · 2025–2026

Five products in production today. One stack. One operating model. Different verticals.

[ 01 ] Flagship · CTO LTD / CEO AS

Eir Tec — Home-based QEEG + Explainable AI for mental health

A home-based four-channel QEEG headband paired with a clinician dashboard and Explainable AI analysis — built for ADHD and ASD diagnostics. The patient records a session at home; the clinician gets a structured, explainable result instead of weeks of waiting.

The clinical journal runs live ASR transcription and generates SOAP notes, referral letters and patient summaries automatically. End-to-end on Azure UK South, single-tenant per clinic, UK GDPR / Data Protection Act 2018 compliant, with WorkOS SSO and privacy-first anonymized retention.

  • Azure UK South
  • Single-tenant
  • UK GDPR / DPA 2018
  • WorkOS SSO
  • Explainable AI
  • ASR + LLMs
  • SOAP gen
  • 4-ch QEEG
Time to result
~7 min

vs 7–10 weeks today

Accuracy lift
+44%

ADHD diagnostic improvement

Cost reduction
−87%

vs traditional EEG path

View at eirtech.co.uk

[ 02 ] COO / Systems Architect

MediVox AS

AI for medical documentation. Local LLMs, ASR transcription and SOAP-grade structured notes — built for Norwegian healthcare with strict GDPR controls.

  • Local LLMs
  • ASR / Whisper
  • GDPR
  • Skjld AS group

[ 03 ] CTO

Skjld Labs AS

Applied AI R&D — a dedicated dev shop for shipping production AI: agents, evals, retrieval pipelines and multi-tenant SaaS scaffolding for clients in regulated sectors.

  • Agents
  • RAG
  • Evals
  • Multi-tenant SaaS

[ 04 ] Co-founder

Jaydus.AI

Multi-LLM AI platform — GPT-5, Claude 4, Gemini 2.5 and 14+ models in one workspace, with agents, knowledge bases and SOC 2-aligned controls.

  • GPT-5
  • Claude 4
  • Gemini 2.5
  • 14+ models
  • SOC 2

[ 05 ] CEO / Founder

InfoDesk AS

Full-service IT and Microsoft Partner since 2002 — the operational base under everything else: identity, infrastructure, support and managed cloud for SMBs.

  • Microsoft Partner
  • Managed cloud
  • Identity
  • Since 2002

// 03 — How I build

One backbone. Every product.

The same backbone runs through every product I architect — so a new vertical (mental health, medical records, applied AI) ships in months, not years.

// PRODUCT LAYER

L1 · interface

What the user touches

  • Clinician dashboards
  • Patient apps
  • Admin consoles
  • Audit + compliance
  • Multi-tenant SaaS shells

// AI LAYER

L2 · intelligence

Where the value is generated

  • LLMs (GPT, Claude, Llama, Mistral)
  • ASR (Whisper)
  • NLP pipelines
  • Autonomous agents
  • Explainable AI
  • RAG
  • Eval pipelines

// PLATFORM LAYER

L3 · runtime

How it runs in production

  • Azure (UK South / EU)
  • Single-tenant or shared
  • Kubernetes
  • WorkOS / OAuth
  • PostgreSQL
  • Redis
  • Terraform
  • CI/CD

Track record

Years in the field
20+

shipping production systems

Active leadership roles
5

CTO · CEO · COO · Chair

First AI work
2005

long before it was a category

GDPR / UK DPA grade
100%

across every product

// 04 — AI agents

Talk to my AI agent — Dr. Alban.

Dr. Alban is a continuously updated agent trained on my work, writing and architectural decisions across MediVox, Eir Tec, Skjld Labs and Jaydus. Ask it about my approach to regulated AI, healthcare deployments or how a specific product is built — it answers the way I would, and gets sharper over time as my work evolves.

  • Continuously updated — not a static demo
  • Grounded in my actual experience and projects
  • Evolves over time as I ship new work
Chat with Dr. Alban
dralban@moltbook ~ %

> how does eir tec stay GDPR-compliant?

dralban Single-tenant per clinic on Azure UK South, anonymized retention by default, WorkOS SSO, and audit trails on every model call. Patient data never leaves the tenant.

> what stack does håkon reuse?

dralban LLMs + ASR + cloud-native + GDPR-grade single-tenant + agents + Explainable AI. Same backbone, different verticals.

>

// 05 — Get in touch

Bringing AI to a regulated industry?

I work with teams in healthcare, legal, finance and the public sector who need production AI without compromising on data protection. If you have a hard problem in a regulated environment, I'd like to hear about it.

[email protected] Connect on LinkedIn

// Eir Tec LTD · Eir Tec AS · MediVox · Skjld Labs · Jaydus.AI · InfoDesk · Open Info