databag Academy · next cohorts open for registration

Your AI agency
from idea to ownership

databag helps companies and people in the Netherlands and beyond transform with AI. We build AI agents and products, run hands-on workshops and trainings, ship sovereign self-hosted platforms, and grow an open community of AI builders.

  • AI Transformation
  • AI Agents & Products
  • AI Learning
  • AI Community

Model-agnostic by design. We build on frontier APIs and on open-weight models you can run on your own hardware.

Frontier APIs

OpenAI GPT
Anthropic Claude
Google Gemini
Mistral Large
Cohere Command
OpenAI GPT
Anthropic Claude
Google Gemini
Mistral Large
Cohere Command

Open-weight · self-hostable

Llama
Mistral
Qwen
DeepSeek
Gemma
gpt-oss
Phi
Kimi K2
GLM
MiniMax
Nemotron
Granite
OLMo
Llama
Mistral
Qwen
DeepSeek
Gemma
gpt-oss
Phi
Kimi K2
GLM
MiniMax
Nemotron
Granite
OLMo

Flagship products

We build what we believe in

Our client work is backed by our own products — battle-tested in production and shaped by real customers.

Live

EarthEmission

Climate tech, powered by AI

Measure, understand and reduce your carbon emissions. EarthEmission turns messy operational data into clear emission insights and reduction plans.

  • Automated emission data collection
  • AI-assisted reporting
  • Reduction roadmaps
Coming soon

Agents Router

One gateway for all your agents

Route every request to the right agent and model, enforce security policies, and observe cost and quality in one place — self-hostable from day one.

  • Model & agent routing
  • Policy guardrails and audit logs
  • Runs in your own cloud or on-prem
Upcoming cohorts

databag Academy

Live, hands-on AI cohorts

Intensive weekend cohorts like the Forward Deployed Engineer program: prototype in hours, ship real AI systems, and graduate with a databag certificate.

  • 5 live Saturdays · 40 hours
  • Recordings + capstone project
  • Company invoice on request

Sovereign AI

Own your agents. Own your data.

We specialise in self-hosted, sovereign agent platforms that run locally in sandboxed environments — secure enough for regulated industries, simple enough for a small team.

Runs where you want

Your laptop, your servers, your private or EU cloud. Open-weight models served locally — no data leaving the building unless you say so.

Sandboxed by default

Agents execute code and tools inside isolated containers with network and filesystem limits, so one bad step can never touch production.

Agent security built in

Prompt-injection defences, least-privilege credentials, human approval for risky actions and continuous red-teaming.

Every action audited

Full traces of what each agent saw, decided and did — ready for your security team and for EU AI Act documentation.

No lock-in

Open-source building blocks and swappable models. If you ever leave us, your agents and data come with you.

GDPR-first

Designed in the Netherlands with privacy by design, data minimisation and clear data-processing agreements.

Frequently Asked Questions

Still have questions? Contact us

Ready to make AI work for you?

Tell us where you are — a first idea, a team that needs training, or an enterprise rollout — and we will propose a concrete next step.