About seven hundred Luxembourg bills see activity in a year — roughly thirteen a week. Here is one of them, quoted in full.
Bodane watches Luxembourg legislation for the people it lands on — fiduciaires, law firms, fund managers and the compliance functions of the financial sector — and warns you a bill is moving through your practice while there is still time to act on it.
Every figure below is counted off the corpus Bodane has already ingested — the whole register of Luxembourg bills and grand-ducal regulations, not a sample of it.
Search all 7,753 titles for blanchiment and thirty-two bills come back. Thirty-two. The bill at the top of this page is one of them, and only because it happens to cite the 2004 law by name. Everything that reaches your obligations through a law that does not use the word is invisible to a keyword filter — which is most of it.
Those thirty-two are also not the same thirty-two you want. Some merely mention the word in passing. Eighteen bills actually amend the 2004 law — and that is a question a keyword cannot ask, because the thing you care about is not a word. It is a law.
You describe a topic — the laws your practice turns on, the European instruments you have to live with, the ministries that own them, and the words that matter when nothing else fits. Bodane sorts every bill in the corpus against it and separates what it is confident about from what it wants you to look at.
The citation buried in an amendment title is an identifier, and Bodane reads it. 1,658 distinct Luxembourg acts are named across the register this way, and 4,188 of the 7,753 bills cite at least one. You subscribe to the law — the loi du 12 novembre 2004, the Code du travail — and every bill that touches it arrives with a reason you can check against the title in a glance: amends loi du 12 novembre 2004.
On the four topics in our own demo account, this turned 169 rows that had been a question into an answer — bills the previous matching could only offer as “review this” or “possibly relevant”, now settled exactly. That is the whole of the gain, and it is worth being precise about what it is: not more bills found, but far fewer left for a person to adjudicate.
A partner does not think in topics. Every firm keeps a list of the laws its mandates depend on, and the question partners actually ask each other is what is happening to the 1967 law. So each act has its own page: every bill in the corpus that touches it, in date order, with what each one did and where it got to.
The loi du 4 décembre 1967 — income tax — has 203 such bills behind it. You can follow an act from its own page in one click, and the page is keyed to the full citation rather than the date, because loi du 22 mars 2004 names three unrelated laws.
A consolidated code is the exception: the Code du travail is the most-cited act in the register at 208 bills, and it can be followed like any other, but it carries no act identifier of its own, so it has no page yet.
A directive becomes Luxembourg law and the register records the link. A regulation — EMIR, the CRR, the GDPR — applies directly and is never transposed at all, so there is no link, and anything matching on transposition is blind to it. Luxembourg still legislates around one: naming the supervisor, setting the penalties, repealing what it replaces.
161 bills in the corpus cite an EU regulation by number, across 194 distinct regulations. Bodane reads those citations, so a regulation can be followed the same way a directive is.
Two independent signals agreed: the topic’s own directive was transposed by this bill, or a keyword in the title lined up with the sponsoring ministry. These are the rows to act on, and there are usually few of them.
One signal, not two — a title keyword hit that no sponsoring ministry agreed with. Ranked by how rare the keyword is across the whole corpus, because a rare term is worth more than a common one.
You mark each one relevant or not. Those verdicts are kept, and they are the labelled set the next tier of matching is calibrated against. Reviewing is not busywork you do for the product; it is how the product learns your practice.
Bills from a ministry you watch, with no keyword agreement. Kept behind a fold and labelled as what it is: a filter, not evidence. A ministry is far too coarse a signal to raise an alert on, and pretending otherwise would bury the four rows that matter under a thousand that do not.
A firm’s templates and engagement letters do not rest on a law. They rest on a provision of it — l’article 5, paragraphe 2, de la loi modifiée du 12 novembre 2004 — and when a bill moves, the question a partner actually has is not whether the law is touched but whether that article is. Bodane now reads citations at that level, on both sides: in the bill’s text, and in documents you give it.
Three steps, and the first one is the only work: you add a document, you read back what Bodane found in it and untick anything that is not right, and from then on you are told. A bill that reaches one of those articles names the document in its own list, in the monthly briefing and in the digest that arrives by e-mail — and opens the article it changed, with the change marked.
You point Bodane at your own template clauses and engagement letters. It reads which provisions they rest on — article by article, not just which law — and you confirm that list before anything else happens. From then on, when a bill amends one of those articles, the alert names the article and the document: this bill amends article 5 of the loi du 6 avril 2013, which this engagement letter rests on — one line per document and article, so a bill that reaches four templates shows four.
It does not wait on a page, either. A bill that reaches an article one of your documents rests on appears in the monthly briefing and in the digest — ranked above the bills that merely became law, because it names work rather than reporting a state — and it can put a bill in front of you that none of your topics caught, which is the case this whole feature exists for. A bill that only names the act, without saying which article, is listed with your alerts rather than sent to you: there are many of those, and a briefing full of them would teach you to skim it.
Three tiers, ranked, and none discarded. The bill amends the article; the bill merely mentions it; or the bill’s title names the act and its text does not say which of the act’s articles it amends. The weak tiers are kept because here the expensive failure is the opposite of the usual one: a missed amendment means a template goes stale silently, which is the thing you are paying to be told about.
Amended and mentioned are different things, and telling them apart matters more than it sounds. Across 20,712 article-level citations in the corpus, 4,438 — about one in five — amend the article they name; the rest refer to it. An alert keyed on mentioned would fire 4.7 times as often as one keyed on amended, which is how a warning becomes noise. Most amendments do not name their law at all: they say de la même loi, or list bare articles under la loi … est modifiée comme suit. Bodane reads those too, and says when the law was read from the text around the article rather than from the citation itself.
A firm’s drafting is privileged and commercially sensitive, and Bodane has no use for it once the citations have been read out. So the text is not stored — true of your template library, which is what this section describes. It is not true of a prospectus you hand Bodane to plan an amendment to: there, what Bodane has read out of the document’s structure is held for as long as that amendment stays open, and let go on the nightly sweep after you finish it — or at once if you remove it yourself, and in any case thirty days after you started it. What is kept from your template library is the title you give each document, a fingerprint that recognises the same file twice, the list of provisions it rests on and, for each one, the single sentence the citation was read from — so that the confirmation screen shows you words you can check rather than a reference you would click through. Nothing from your documents is ever sent to the provider that reads bills.
Inside your own walls, the documents never leave your network at all. That is the strongest version of the argument for self-hosting, and this feature was built so that it could be made.
An alert that says article 5 changed, and nothing more, sends you to the register to find out how. So when a bill amends an article, Bodane puts the article as it now reads beside the article as it read before, with the difference marked word by word — because a Luxembourg amendment is typically three words in a paragraph of several hundred, and highlighting the paragraph hides the change inside it.
The State publishes the consolidated text of each act with the date each version began to apply, and the comparison is built on that. Which two versions to compare is decided by that date, never by the date of publication: consolidations are published weeks after the amendments they embody, and a bill’s changes often commence in stages.
Opened from an alert, the comparison is the change that bill made — the versions either side of the day it came into force. A bill not yet in force has changed nothing in the consolidated text, and the page says so and shows the article as it reads today, rather than somebody else’s amendment.
What has been measured, and what has not. The reading of citations was measured on 4,236 legislative documents — 238 million characters of Luxembourg legal French — before the feature was built. 13,007 citations bound an article to a dated act; 53% of documents carry at least one; 99.6% resolve to an act Bodane already holds. A random sample of 22 was read by hand and every one was correct, which is a small sample and is stated as one. Recounted in October 2026 with what the reading covers now — la même loi, listed amendments, hyphenated article numbers — the same corpus yields 20,712 citations, and two further samples of inferred amendments, 30 and 22 drawn at random, were read by hand without a wrong one.
It has not been measured on a firm’s templates. Contracts may cite in English, abbreviate to art. 5, or follow house conventions the register never uses, so the share of your own citations it reads is unknown until it has read one firm’s documents — which never have to leave that firm’s network for it.
A Luxembourg bill rarely arrives alone. It usually comes with a statement of reasons, an impact assessment and an article-by-article commentary — documents written precisely to explain what the title does not. They are public, they are attached to the dossier in the register, and nobody in a fiduciaire has time to read thirteen sets of them a week.
Reading those documents, and matching your topics against what a bill actually does rather than what it is called, is what Bodane is for.
It has started. 380 bills have been read end to end, and every one of the 3,540 subject claims kept was located verbatim in our own stored copy of the text — a quote we cannot find is discarded rather than shown. That is 15% of the 2,528 bills carrying readable text, so treat it as a bridgehead. Nothing else on this page depends on it.
Why the bill exists, and what its authors say it will change. 743 of them carry readable text.
Who it lands on, and what it is expected to cost them. Its two useful fields are extracted on our own machines and the document is on a never-leaves list: 86% of first pages carry a named official’s direct line and email.
Article by article, what the text of the law becomes. 712 of them carry readable text.
For a transposition, which article of the directive becomes which article of Luxembourg law. Only 75 exist with readable text, and they are not in the set sent for reading.
These are the events the register actually records against a bill, in its own words. A tool that watches the Journal officiel speaks at the last one. By then the text is fixed, the consultation is closed, and the only thing left to do is comply.
How long you actually have. Across the 2,878 projets de loi in the register that reached the statute book, the median run from tabling to enactment is 239 days — the middle half between 134 and 400. A grand-ducal regulation is far quicker: median 79 days, over 376 of them. Bodane shows a bill’s own elapsed time against the bills it is comparable to, so “this is taking a while” is a measurement rather than a feeling.
Comparable bills means bills that finished. One still in progress is not in that comparison, so an unusual figure means unusual — never overdue.
Withdrawals count too. A bill you prepared for can be pulled — retrait — and that is worth knowing as early as anything else. Bodane records it as an event like any other rather than letting the dossier go quiet.
And the dates that are still ahead. A transposition deadline set in Brussels, a commencement date published by the Grand Duchy, and a date written into the body of a bill arrive from three unrelated places. Bodane puts all three on one forward list, and issues a private address you paste into Outlook, Google or Apple Calendar — so they sit beside client meetings rather than on a page somebody has to remember to open. With the e-mailed digest, it is how the product reaches a partner who never signs in.
The list is not called deadlines, and the wording is deliberate. The corpus contains au plus tard le 10 juillet 2027, a real compliance deadline, next to trois arbres après le 1er janvier 2030. Each row shows the sentence its date came from and names the source: a date read out of a bill’s text is labelled as a machine reading, and only the dates the EU and the Grand Duchy publish as such are presented as official.
Event names reproduced from Légilux, Service central de législation, Ministère d’État — licensed CC BY 4.0. English glosses are ours.
The hard part of a supervisory conversation is rarely the bill you missed. It is showing that you were looking at all — on which dates, for what, and what you concluded.
Every decision Bodane reaches is written down and never altered afterwards: which topic, which bill, what it concluded, the reason it gave, the date, and — where somebody reviewed it — who, and what they decided. That record accumulates whether or not anyone asks for it. You can download the whole of it as a spreadsheet.
Including the thousands it assessed and judged too weak to raise. Those rows are the point. A list of what you found proves what you found; a register that also shows what you looked at and set aside is what shows the watch was running.
To be exact about it: the register holds every decision that carried a signal, weak ones included. A bill that matched a topic on nothing at all is not written down — that would be one row per topic per bill — so it evidences what was assessed and judged, not that every bill was individually considered.
What a row says. The topic, the revision of that topic that decided it, the date, the verdict, the rule that reached it, the reason in the words a reviewer saw, the bill and its link back to Légilux, and the human verdict if there is one.
“Not yet reviewed” and “reviewed, not relevant” are kept apart. Collapsing them would close an item nobody had looked at.
A draft note for the client, in French. Knowing about a bill is half the job; telling forty clients about it is the other half. For any bill that has been read, Bodane assembles a draft note — what the bill does, who it lands on, which dates matter — that a fiduciaire edits, puts their own name on, and sends.
Assembled, not written fresh: every sentence of substance comes from the machine reading of that bill’s own documents and is stored with the quote it came from, so it can be checked against the text before it is sent. The note is in French because the client reads it; the product itself can be used in French or in English. It is marked a draft, on the page and in the file, and it is not advice.
And something between the digests, when it earns it. A weekly digest is right for almost everything. Two events are not: a bill you follow being voted, and a bill you follow being published. Those go out on their own the next morning, capped at eight bills — because an alert that arrives every week is a digest with a worse name.
Where Bodane runs is a choice you make, not a tier you buy. The product is the same either way, the price is the same either way, and inside your own walls you also decide whether the two steps that leave the EU run at all. Run that way, any templates you point Bodane at never leave your network.
You evaluate vendors for a living. Here is the list you would otherwise have to extract from us.
No logos, no testimonials, no “trusted by”, no adoption figure — because none of those would tell you whether the thing works on your practice. What the site offers instead is counted: every figure comes from the ingested register and can be recounted, and the demo runs on that same register rather than on a prepared example. The demo shows a fixed snapshot of that register, taken when it was last updated. It fetches no new bills and sends no digests or alerts; a topic you define there is matched against the snapshot as you save it, and a document you add to the clause library is read, but raises no alerts until the register moves — which on the demo it does not.
What runs today matches on titles, sponsoring ministry, bill type, EU-directive transposition, and the acts and European regulations a title cites. All of those are identifiers — exact, checkable, and no substitute for reading. Reading the exposé des motifs and the commentaire des articles has run over 15% of the bills that carry readable text, and a match made that way is shown as review this rather than as a settled answer: those matches have not yet been calibrated against a practitioner’s verdicts, so we do not present them as ones you can rely on. Anywhere the two tiers are mixed on this site, they are labelled.
Naming a law does not surface bills a keyword would have missed — on the money-laundering law it is one bill in eighteen. What it changes is certainty: an exact reason instead of a word that happened to appear, and 169 rows in our own demo account that stopped needing a person to decide them. It is a precision gain, and we would rather say so than let it be read as a coverage one.
Bodane finds regulations in the text of bill titles, so for most of them it can offer one only when Luxembourg has already legislated about it at least once — it will tell you about the second bill concerning a regulation, and cannot warn you about the first. Six are the exception, chosen by hand because the first bill is the one a firm needs: AMLR, AMLA, the transfer-of-funds regulation, DORA, MiCA and the GDPR can be followed before any bill mentions them. A directive does not have that limit, because the register publishes the transposition link directly.
The reading of article-level citations was measured on 4,236 legislative documents and a hand-read sample of 22, every one correct. That establishes that it works on Luxembourg legal French as the register writes it. It does not establish what share of a firm’s own citations it reads — contracts abbreviate, cite in English and follow house conventions — and until one firm’s documents have been read, no such figure is given. The feature is built, tested and on the demo.
Averholm’s other product does, and can. Bodane sends redacted public legislative text to a provider in the United States, because no EU processing region is available to it. Client data never goes near that step; the one thing of a firm’s that can is a description of its practice, if it chooses to write one when setting up. The whole boundary is written out.
The corpus figures on this page come from the ingested database in September 2026 and can be recounted. Where a figure would have to be guessed, there is no figure — which is why this site quotes no time saved and no coverage percentage, and the one accuracy figure it gives is a hand-read sample of 22, stated as such.
We will walk you through Bodane on the real Luxembourg register, in the areas your own practice turns on. If it looks useful you can go on to define those topics properly and mark what comes back — which is what shapes the product, and how you see it first.