Delfen
Practitioner Platform · AI Era

The AI era doesn’t replace the IT practitioner.

It makes you more essential.

If you know how to lead it — not just use it.

AI can now write the Terraform, draft the policy, and generate the architecture diagram in seconds. It cannot decide which of those to trust. That decision is the job — and it is the one thing AI makes more valuable, not less. Delfen.ai works through those decisions with 30+ years of practice behind it — infrastructure engineer to IT Director and Chief Enterprise Architect, much of it in DNB-regulated financial services. Named tools. The failure modes. A verdict you can act on.

Featured: open source

Sovereign Agent Logbook

Every AI agent needs a logbook. Yours should stay in Europe.

The first question after an agent goes wrong is what it did and who approved it. This logbook answers that from your own server: each action, tool call and human approval the agent logs, kept in a tamper-evident file. Works with any model. One Docker container, no outbound network, Apache 2.0.

  • Tamper-evident: change, insert, remove or reorder a line and the verify command tells you exactly where
  • Any model, any cloud, or no cloud: built on MCP, the open standard for connecting agents to tools
  • Audit-ready: who, what, why, approved by whom, with what outcome, exported to JSON or CSV

v0.1: early, working, tests passing. Hosted version on EU servers, reserve now, not yet live: €29/month, 30-day free trial.

Illustrative output, abbreviated

$ logbook verify

OK — 212 events, chain intact

$ logbook tail --limit 1

{

"seq": 212,

"agent_id": "invoice-bot",

"model": "mistral-small-4",

"action": "pay_invoice",

"arguments": { "amount": 1250, "iban": "***masked***" },

"approval": "auto",

"approval_rule": "pay-under-2000-matched-po",

"outcome": "success",

"prev_hash": "9f2c…", "hash": "3a1b…"

}

Why this platform exists

Most AI strategy content is written for boards.

The frameworks are abstract. The tools are unnamed. The failure modes are never mentioned. And the people doing the actual work, the engineers, architects, and technology leaders making decisions every day, get a summary PDF and a keynote slide.

Delfen.ai is for the people doing the work.

Real HOW, not fake HOW.

Named tools. Specific configurations. Verdicts you can act on. Not "consider implementing an AI strategy."

From practice, not from research.

30 years making architecture decisions that shipped — at Fugro, NIBC, and Van Oord, not in a research paper. The judgment is earned, not synthesized.

Independent, and open about it.

Where an affiliate relationship exists, the article discloses it. The verdict is written first, and the disclosure never changes it.

The book series

The 10x Enterprise Architect with AI

The 10x Enterprise Architect with AI

AI creates. The architect leads.

Jacques Domenie · Delfen.ai

Writing in progress

The 10x Enterprise Architect with AI

If you can’t beat them, lead them.

AI creates. The architect leads. The first book in the 10x Enterprise Architect with AI series — spanning all four architecture domains, for architects navigating the AI era without losing what makes them essential.

Not a beginner’s guide to AI. Not an executive summary. A practitioner’s playbook for the enterprise architect already deep in the work. It covers what changes, what stays the same, and how to lead a team through both.

In preparationDelfen.ai
Join the waitlist →

Delfen CTO update

What changed for a decision you are making.
Every Tuesday.

For CTOs, CIOs and board members in financial and industrial organisations. Three to five developments a week, each tied to a decision you make, each with a question to forward to your team.

Free. Independent. One email a week. Unsubscribe anytime.

0

decision per article, worked to a verdict

0+

years of practice behind every verdict

0%

of article claims fact-checked before publishing