· FAQ
Frequently asked questions.
Concrete answers without the sales pitch. Missing something? Ask us.
Local AI and technology
What is local AI?
Local AI is artificial intelligence that runs on hardware you control, not in a third party's cloud. That means data never leaves the EU jurisdiction. We primarily use open-source models like Gemma, Llama, and Mistral, hosted on Tenki's Mac Mini or your own infrastructure.
When is local AI worth it over cloud APIs?
When the data can't leave the EU (health, municipal, legal, finance), when latency must be predictable, when cost must be predictable, or when you need to answer for what the model was trained on. For most other cases, cloud APIs from Anthropic or OpenAI are cheaper and faster — we won't sell you local AI if it doesn't actually help you.
Which models do Hugin and Munin use?
Hugin runs Gemma 3 on Tenki's Mac Mini in Oslo — everything is processed locally. Munin calls cloud models (typically Claude or Gemini) anonymised. Which one is active is shown clearly at tenki.no/chat.
AI Act and GDPR
When does the AI Act start to apply to me?
The EU AI Act entered into force on 1 August 2024. The most important parts for most businesses are enforced from August 2026. If you use AI in customer contact, decision-making, or internal operations, you should start mapping it out now. See /en/ai-act for details.
What is the difference between GDPR and the AI Act?
GDPR is about personal data and how you process it. The AI Act is about AI systems' risk and documentation — regardless of whether the system processes personal data. Many AI systems are covered by BOTH GDPR and the AI Act, and the documentation has to reflect both.
Do I need a DPIA (Data Protection Impact Assessment)?
Probably yes, if you use AI for higher-risk processing of personal data (recruitment, credit scoring, health data). Tenki delivers ready-made templates as part of the AI Act assessment. We also have an open data processor agreement template to build on.
Pricing and collaboration
What does an AI assessment cost?
From NOK 25,000 fixed price. Includes interviews with key people, mapping of existing AI use, identification of GDPR and AI Act risks, a written report with prioritised actions, and a 30-minute presentation of the findings. See /en/services.
How long does a typical engagement take?
AI assessment: 2–4 weeks from signed agreement to delivered report. Implementation projects: variable scope, we give a concrete estimate after the assessment. We only take on what we can deliver well, not just what fits the calendar.
Do you take small jobs, or only long engagements?
Both. The AI assessment is deliberately designed as a low- threshold entry (3–4 weeks, fixed price). If you just need a sanity check on a specific AI implementation or a legal review, email us — we'll say no if it isn't Tenki's niche.
What aren't you good at?
We aren't lawyers. We don't give final legal advice on complicated AI Act questions — we help you prepare the material and suggest when you should consult an attorney. We also don't do generic IT consultancy (Microsoft 365, phone systems, etc.) — we are built specifically around AI.
Tenki as a company
How large is Tenki?
A bootstrapped AI startup. Built and run by Einar K. Holt in Oslo, with summer interns when there's capacity (/en/careers). We plan to grow carefully — only when client work demands it.
Are you Norwegian-owned and Norwegian-operated?
Yes. Tenki Labs AS is registered in Norway (org. nr. published when assigned by Brønnøysund). All our main infrastructure runs on servers in Norway or the EU region.
Where can I find more detailed technical content?
/publiseringer has blog and research articles (Norwegian). Our GitHub organisation is github.com/tenki-labs — there you'll find, among other things, NORI (Norwegian LLM evaluation) and kibarometer (Norwegian labour-market data).
Your question wasn't here?
Send a line to [email protected] or take the maturity test as a starting point.