The third tier. It requires Operations and Cherub — it is not available on its own.
Cherub looks inward, at your code. Cassandra looks outward: at a list of sources that you lay down. If something changes there that might affect your application, you get an alert — in plain language, with the source, so you can read it up yourself. What follows from it is your decision.
Watch
Cassandra follows the sources on your list: official publications and authority websites on the topics that concern you, the vendors of the building blocks your application runs on, and the part of your market that you name.
Alert
If something changes there, you get an alert. One sentence in plain language, the source so you can read it up, and a pointer to where something like that occurs in your application at all.
Implement
If you decide that something has to change, we build it — once you approve. Small things go through in ongoing operations, bigger ones become a project of their own with their own plan and a fixed price.
And the fourth one:
Whether an obligation applies to you is not something we judge. That is a legal question, and your lawyer, your tax adviser or your chamber answers it. Nor are we allowed to: legal advice is a regulated profession — in Germany under the Rechtsdienstleistungsgesetz, the Legal Services Act, in Austria under the Rechtsanwaltsordnung, the code of professional conduct for lawyers. That is not a cautionary line at the bottom of the page, it is the boundary this tier is built along.
From public sources. Official publications, authority websites, security and release notices from the vendors, publicly available market information. No law firm behind it, no legal information, no insider knowledge and no access that you would not have as well. What Cassandra takes off your hands is the searching and the matching up — not the knowing.
The list takes shape in the intro call and is in writing before the tier starts: which sources, which topics, which areas of your application. That way you see both — what we follow and what you have to keep an eye on yourself. If a gap turns up later, we take it in. From then on, not retroactively.
This is what a list can look like. Yours looks different.
- The rules on invoicing public sector clients in your country.
- The vendors of the building blocks your application runs on — new versions, security notices, announced end-of-life dates.
- The standard or the marking you produce to.
- The interface of a client that your application hangs on.
- Two or three competitors whose offering you want kept an eye on.
An alert is a pointer, not an instruction. You read what has changed, where the information comes from and which part of your application might be touched by it — with the source, so you can read it up yourself or put it in front of your lawyer or tax adviser. Whether an obligation applies to you is not something we judge; we are not allowed to either. If you decide that something has to change, you tell us — whatever follows from it technically, we build once you approve it. Small things go through in ongoing operations, bigger ones become a project of their own with their own plan and a fixed price. If you do nothing, nothing happens: an alert sets no deadline and changes nothing in your application by itself.
01What has changed
One sentence in plain language. No section numbers, no abbreviations.
02Where we have it from
Source, where it says so, from when. You can read it up without asking us.
03Where something like that occurs in your application at all
Which part, which building block. That is a technical assignment, not a legal one: we tell you where it sits in your system — not that the rule applies to you. If it is about the market and not about the code, nothing stands here.
04What could be done technically
A proposal with a rough order of magnitude, non-binding. Not a quote and not an order — nothing gets built until you approve it.
05What we do not say
Whether it applies to you. That is a legal question, and your lawyer or your chamber answers it, not us. And what we do not know goes in as a gap, not as a guess.
That is how an alert is built. How many there will be in a month we do not know ourselves yet — we do not have customers yet. What is promised is the form, not the quantity.
- What has changed
- For invoices to public sector clients, a different exchange format applies from a cut-off date on.
- Where from
- The official publication, with citation and date. In the real alert it stands here — to read up, not to take on trust.
- What that has to do with your application
- Your invoice export. Today it produces a PDF; what is required is a structured file.
- What would be possible technically
- A second export format alongside the existing one, roughly two working days on a first estimate. The current export stays as it is. It only gets built when you commission it.
- What we do not say
- Whether the rule applies to you. That depends on who you write invoices to — and that is for your tax adviser or your lawyer to judge, not us.
Cassandra cannot reach your code or your data. It watches what happens outside and alerts you to it — change anything it cannot. From the alert to the change it is four steps, and three of them are yours: it alerts. You decide whether it concerns you. You decide whether we change anything. And it only goes live once you approve it.
Watching is only of use to you if you know where it stops. Otherwise you rely on it at a point nobody ever looked at.
- No deadline monitoring. Cassandra alerts you that something is changing — it does not keep a calendar of your obligations and it does not remind you of any filing date. Reminders like that can be built into your application, but they are not Cassandra’s job. Keeping an eye on the deadline stays with you.
- No wording that has to hold up legally. What applies in your name does not come from us: contract clauses, the privacy policy, the letter to an authority. Not in the small places either — the consent sentence under the form, the retention period in the deletion routine. There we ask you for the wording instead of inventing one. You decide what it says. We build it in.
- No completeness. Cassandra finds what is in public sources. It can miss something, and it does not replace professional advice. If you did not get it, you did not get it — which is why what gets watched is in writing.
- No promise about how often. How many alerts add up depends on your trade and on your list. We do not invent a number for that — and we do not turn a quiet month into an alert so the tier looks busy.
- No substitute for your own diligence. You stay responsible for your business, and that responsibility cannot be bought in — no more from us than from anyone else. Cassandra takes the searching off your hands, not the judging. Anyone who sits back because somebody is looking has bought the wrong tool. What turns up lands on your table — what becomes of it is your decision.
If too little is left over for you after that list, leave tier 3 out. Operations and Cherub run just the same without it.
- What if you have missed something?
Then you did not get it. That is why what gets watched is in writing — which sources, which topics, which areas. You get the list before the tier starts: so that you see both, what we follow and what you have to keep an eye on yourself. If a gap turns up, we take it in — from then on, not retroactively.
- Can’t a search alert on the web do that?
In part, yes. A search alert tells you that a word came up somewhere. Cassandra tells you that something has changed at a named source — and adds where something like that occurs in your application and what could be done about it technically. The difference is not the finding, it is the matching up, and that afterwards there is somebody who builds it in.
If a search alert is enough for you: take it. It costs nothing.
- Why can’t I book Cassandra on its own?
Because an alert is only worth something once somebody knows where it belongs in your application — and because the change that follows from it needs somebody to build it. Both of those are tier 1 and tier 2. Cassandra on its own would be a circular.
When tier 3 is worth it
- You supply public sector clients, or corporations that pass their requirements down the chain.
- Your application hangs on an interface somebody else runs — and they announce changes when it suits them, not when it suits you.
- You produce to a standard or a marking that can change.
- You have found out about a change too late once before and still remember what it cost.
When you do not need Cassandra
- Your application covers an internal workflow that nothing outside hangs on.
- Your tax adviser, your chamber or your trade association keeps you informed anyway — and you read what comes from there.
- You keep an eye on the market yourself, because you know it better than any source does. That happens, and when it does it is usually right.
In place
- Cassandra is built and running — on our own software.
- Following the sources on a list and spotting changes to them.
- Building an alert in the form set out above, with source and date.
- The assignment to a place in your application, where it is about code and not about the market.
Not in place yet
- Experience of how many alerts add up in which trade — that takes customers, and we do not have any yet.
- A statement about what share of the alerts turns out to be usable in the end. That one only comes with you.
- A ready-made list of sources for your trade. It takes shape in conversation, not off the shelf.
What would be on your list, we do not know yet
That is the real work in tier 3, and it happens in conversation. Tell us where the unpleasant surprises have come from so far — a client, an authority, a vendor that discontinued something. The list comes out of that. It is in writing before the tier starts. And if there turns out to be too little on it, we tell you so.
Look at the technology first