Зміст

From the outside, digitalising a court system looks like big projects: registers, subsystems, specifications worth millions. Yet a large share of the actual effect comes from somewhere else entirely — changes to how an institution runs internally, requiring neither procurement nor new legislation.

What follows is an overview of proposals submitted as internal memos: what was proposed, what it achieved, and where it stands. Not a retelling of the documents themselves — what matters is the problem each one addresses.

Paper records: the most expensive habit

The easiest strand to grasp, and the most revealing.

The judicial governance body has a dedicated electronic records system. It is a complete one: registration, search, drafting procedural documents, tracking execution, transfer to the electronic archive. The problem was never a missing tool. The problem was that only part of it was used, in parallel with paper.

The cost of running both shows up in the budget. Public procurement gives the scale: postal services, servicing and repairing office equipment, paper — hundreds of thousands of hryvnias a year on processes that could largely happen without paper at all.

The effect of changing approach was measured in practice. Preparing requests electronically saves the registry office 10–15 minutes per document: nothing to print, no paper copy to register, no dispatch to arrange. In one month, more than eight hundred such requests were prepared — roughly 136 hours, or seventeen working days of the records unit.

What was proposed: bring the records instruction into line with what the existing system can already do, drop the mandatory printing of certain categories of document, move the signing of decisions to qualified electronic signatures, and extend electronic handling of case materials.

One argument that has nothing to do with savings: electronic records are institutional resilience. A body that works with paper originals in one building stops when that building stops. In a city under regular shelling, that is not a hypothetical.

Status: one disciplinary chamber now signs all its decisions with a qualified electronic signature. The proposal was to extend that practice across the whole body.

An analytics unit: capability, not access

The right to access state registers does not, by itself, create the ability to use them. A few dozen registers you can reach, with nobody able to cross-reference data between them, amount to a list of links.

Hence a separate strand: setting up an analytics unit that collects information centrally, works with resources requiring secure workstations, cross-references data from different sources, and prepares material for decisions.

The full package was worked through: the unit’s charter, job descriptions for the head and senior specialists, research methodology, procedures for framing analytical tasks, and training materials on working with data and open sources.

The value is not the unit itself but the centralisation. The same work used to be done by members of the body, their assistants and disciplinary inspectors — each for themselves, in parallel, duplicating one another. Centralising it freed up dozens of people for what they are actually there to do: legal assessment and drafting decisions, rather than copying data between files.

An honest caveat: concentrating a large volume of manual work in one small unit is not an achievement. It is a diagnosis. It shows precisely what should be automated next.

Access to state registers

A strand that took years of correspondence.

The law gives members of the body direct access to state information resources, with no court order and no permission procedure. The wording is clear, and no data controller disputed it.

What they disputed was different: the regulation governing a specific register contains an exhaustive list of authorised users, and the body is not on it. The law grants the right; a lower-ranking instrument does not see it.

This is a standard problem in Ukrainian digitalisation, and it is not technical. Every data controller guards its own register, because a leak is its responsibility while another institution’s speed of work is not. The incentives are arranged so that caution always pays better than cooperation.

What was proposed: aligning secondary legislation with the law, drafting procedures for electronic exchange, obtaining access to the data exchange module, and speeding up the process of granting access.

Status: access to some registers was obtained, and they are now everyday tools. One distinct result: full access to the register of court decisions not only for members of the body but also for secretariat staff and disciplinary inspectors — the people who actually prepare the material.

Keeping disciplinary cases under control

Two connected problems, both solvable with what the records system already had.

Prioritisation. The law and the rules of procedure require cases to be handled according to priority, and define the types of priority — from cases that could end in a judge’s dismissal to those attracting significant public interest. The regulation states plainly that priority is set in the records system. The system had no such function.

The proposed fix is notable for its thrift: rather than commissioning new development, reuse the existing control-card mechanism, adding fields for priority type and target and actual completion dates. Prioritisation is, in substance, a form of tracking.

Deadlines. Disciplinary proceedings are capped at eighteen months, and passing that limit has one consequence: the case is closed. A missed deadline means the case is never heard on its merits, whatever its circumstances.

The proposal: calculate the remaining days automatically, show that figure in the case list, add filters by handler and colour-code by time remaining, with a warning at least six months out.

A problem found along the way: the records system was configured with default settings for everyone. Staff in different units saw identical tables full of technical columns they had no use for. That created a false impression that the software itself was poor, and put people off using it seriously. The proposal was to configure profiles around actual job roles.

Document templates

The template mechanism existed in the records system long before document automation became a fashionable topic. It pulls data from the system into a document, writes entered values back to the database, supports conditional logic, signs the result electronically and files it against the case automatically.

For years, barely anyone used it.

There is a lesson here that goes beyond one institution: the problem in digitalisation is rarely a missing technology. Far more often, nobody owns the technology that is already there as a task in its own right. No process owner, no templates, no training, no requirement to use it.

The result: more than fifty templates for procedural and organisational documents. The clearest example is the standard ruling returning a complaint — preparation went from one or two hours to a few minutes. Gone is the print-sign-scan-upload loop that many institutions still treat as normal.

Someone has to own digital development

Digital projects tend not to survive without a permanent owner. As long as digitalisation is an extra load carried by a few motivated people, it advances at the speed of their spare time.

So several memos proposed creating the post of deputy head of the secretariat for digital development — a CDTO, as other state bodies have.

This does not duplicate the IT department. IT keeps the equipment and systems running. A CDTO owns architecture, priorities, the link between technical change and real business processes, and coordination between institutions, the state enterprise administering the systems, and international partners — so that ten donor-funded projects do not become ten incompatible systems.

Related proposals: naming who is responsible for digitalisation, and for delivering the roadmap.

AI: rules, not a ban

The proposal started from an observation: staff already use these tools, but without any order to it — no regard for what a given model is suited to, and from personal accounts. Banning it would mean pretending otherwise. Leaving it alone would mean accepting the risk.

The proposal was to bring order: define which tasks suit which tool, provide official accounts instead of personal ones, and set rules of use.

Separately, one design decision removes the main objection about sensitive data. The document is anonymised locally, personal data replaced with synthetic values; the anonymised text is what gets analysed; the real values are substituted back afterwards. The model never sees genuine personal data. The anonymisation itself needs no AI — regular expressions and local named-entity recognition handle it.

The principle written into every one of these documents: the final decision is always made by an authorised human being.

What it all points to

Put the proposals side by side and one thing stands out: almost none of them required buying anything.

Prioritisation reuses the existing control-card mechanism. Deadline tracking configures fields that were already there. Templates use a mechanism that had existed for years. Electronic records means actually using a system already paid for. Anonymisation runs on regular expressions.

That is the most accurate diagnosis of digitalisation in the public sector. The losses come not where technology is missing, but where nobody works on the technology already in place: the process is not described, nobody is responsible, roles are not configured, and the written instruction has never been aligned with what the system can do.

The second common thread is that the effect is measurable. Not “improved efficiency”, but specific quantities: seventeen working days a month on one type of document, a standard ruling cut from two hours to a few minutes, the time of dozens of staff freed up.

And the third: most changes began as a pilot in one unit. A practice proven on a small scale is far easier to argue for than a concept. The people who took to the new tools fastest were those who joined after they appeared, with no habit of reaching for paper. That suggests a strategy: unlearning is harder than learning, so every new process should start out paperless.