Every gate.One plan.
Sluyz is the access layer for your data platforms. One set of rules decides who may reach which data, in every system. You see each change before it happens, and you can close everything in one move.
- Works today with Databricks, Azure and PostgreSQL
- Accounts in Microsoft Entra ID
- EU cloud or self-hosted
- Nothing between your people and your data
This is what it looks like in Sluyz. From rule to applied, in four steps.
Rule in UAL
# tagged: payment_detailspolicy finance_read { allow read on finance.invoices to group finance-team}
In plain words
The finance team may read invoices, in Databricks and Azure SQL.
Data tagged as payment details
Rule saved · version 1
- Who
- Finance team · group in Entra ID
- May
- read
- What
- invoices · payment details
- Where
- Databricks, Azure SQL
Reading 5 sources
- Databricksread
- Azure Blob Storageread
- Azure SQLread
- PostgreSQLread
- Microsoft Entra IDread
2 differences found
j.devriesDatabricksNot in any rule. Left in March.svc-legacy-exportEntra IDMember of finance-team. Not in any rule.Tide Plan3 changes
- +Finance team may read invoicesDatabricks
- −Remove j.devriesDatabricks
- −Remove svc-legacy-export from finance-teamEntra ID
Applying
- Databricks
- Databricks
- Entra ID
All sources on level. Access follows your rules.
Sluyz Register · Approved and logged
The rules are yours. Sluyz carries them out.
Sources
5
Databricks, Azure Blob Storage, Azure SQL, PostgreSQL and Entra ID, today.
Rules
1
One rule for every system, instead of one per system.
Emergency stop
2
Two people agree, then every gate closes in one move.
New accounts
0
Sluyz never creates users or groups. It uses the ones you have.
Four products. One model underneath.
Access Map
Who can do what, and where?
Every identity on every platform, with the level it really has: read, write, admin or owner. Including rights through groups and inheritance.
- People, groups and service accounts
- Direct, through groups and inherited
- Access nobody approved turns orange
Illustrative example
Access nobody approved
Register
Who can reach our personal data?
Per category of personal data: who can reach it, on which platform, since when and under which rule. On one page that legal can read.
- Categories tagged by you
- Exports as a dated, signed PDF
- Helps demonstrate GDPR Art. 5(2), 30 and 33
Illustrative example
Register_2026-11-13.pdf · dated and signedTide Plan
Does what is live match what was approved?
Every change is a plan first: what is added, changed or removed. Nothing happens until one or two people approve it.
- Reviewed like a pull request
- Applied natively on each platform
- Checked for drift straight after
Illustrative example
+ grant read · Finance team → Azure SQL / ledger
~ change admin → read · svc-etl → Databricks / sales
− revoke admin · Ex-employee → Azure SQL / ledger
− revoke read · Contractors → Blob Storage / exports
Storm Barrier
Can we close everything, now?
Revoke all human and machine access across every platform and Entra ID in one logged action. Restore the last approved state from the same model.
- Two keys, break-glass stays open
- Everything, or one platform or identity
- Evidence from the first second
All gates closed in one action · logged
One rule. Not four dialects.
The same rule, “finance analysts may read the invoices”, written four times by hand. With Sluyz it is written once, and Sluyz produces each platform’s native form.
Today, by hand
GRANT SELECT ON TABLE prod.finance.invoices TO `finance-analysts`;
GRANT SELECT ON OBJECT::finance.invoices TO [finance-analysts];
GRANT SELECT ON finance.invoices TO finance_analysts;
az role assignment create --role "Storage Blob Data Reader" --assignee-object-id <group> --scope <container>
Once, in UAL
# tagged: payment_detailspolicy finance_read { allow read on finance.invoices to group finance-analysts}
Illustration of the idea, not final syntax.
So access needs one engine. Above every system, not in the way of any.
Who
- People
- Groups
- Service accounts
Outcome
- Grant
- Revoke
- Report
- Close
Sources
- Databricks
- Azure Blob Storage
- Azure SQL
- PostgreSQL
- Microsoft Entra ID
- Snowflake · coming
- Amazon Redshift · coming
- Salesforce · coming
- Microsoft Dynamics · coming
Why can Anna read invoices? Sluyz can tell you, step by step.
The question
Can Anna read the invoices table, and why?
AAnna de Vries · Finance
Sluyz works it out
- 01Find the person
Anna exists in Entra ID. Sluyz does not create her; it finds her.
- 02Follow her groups
Anna is a member of finance-team.
- 03Find the grant
finance-team may read schema finance in Databricks.
- 04Follow the data
The table invoices sits inside schema finance, so the grant covers it.
The answer
Yes, she can read it
- through group finance-team
- passed down from schema finance
- allowed by rule finance_read
Her name is in no grant. That is exactly the access that spreadsheets miss.
A data breach
72 hours to report. Know who had access in minutes.
Most of those 72 hours usually go to finding out who could reach the data. With Sluyz that answer already exists.
Without Sluyz
With Sluyz
Answer from the Register
Storm Barrier
When the storm comes, one barrier closes every gate.
During a hack, Storm Barrier closes all access to all systems in one move, accounts included. Afterwards, the last approved situation is put back the same way.
More about Storm Barrier- DatabricksopenclosedClosed
- Azure Blob StorageopenclosedClosed
- Azure SQLopenclosedClosed
- PostgreSQLopenclosedClosed
- Microsoft Entra IDopenclosedClosed
- Incident teamstays openClosed
Try it: let both people agree, then close everything. This is a demonstration, not connected to a real system.
Two keys
Two authorised people activate it, with a tested runbook.
Break-glass stays open
The incident team never locks itself out.
Choose the scope
Everything, one platform, one dataset or one identity.
Evidence from second one
Every step logged for GDPR and NIS2 reporting.
Dry run
Rehearse it without closing anything.
Built for the questions your auditor asks.
EU hosting
Runs in our cloud in the EU.
Self-hosted
Or inside your own environment.
No proxy
Nothing sits between your people and your data.
Settings only
Only access settings are read and changed, never the data.
Two keys
The emergency stop needs two people.
Full log
Every change has a plan, an approval, a time and a name.
Five design partners. Help us build what comes next.
Sluyz works today and has no customers yet. Partners get a free check of one system, a say in the roadmap and partner pricing.
How it works
From a rule on paper to access that is right everywhere.
Four steps. You stay in control at each one: Sluyz never changes anything without approval.
Step 1: Write the rule once
Rule
The finance team may read invoices.
- Applies to
- Databricks, Azure SQL
- Kind of data
- Payment details
- Written
- Once, for every system
Step 2: Compare with reality
Step 3: See what will change
- Finance team may read invoices
- Remove the access of a colleague who left
- Remove the access of an old supplier login
Step 4: Approve, and it is applied
What stays the same.
Your systems
Nothing is placed between your people and your data. Each system keeps doing its own access control.
Your accounts
Sluyz never creates users or groups. It only decides what the accounts you already have may reach.
Your decisions
You write the rules and you approve the changes. Sluyz carries them out.
Your data
It stays where it is. Only access settings are read and changed.
We are looking for five design partners.
Sluyz works today and has no customers yet. We want to build the next part with a small group of security, legal and data teams.
Ask for a conversationFor legal and compliance
Show who has access to personal data, and why.
Sluyz Register answers the question regulators and auditors ask first. In plain language, with a date, without needing a technician.
| Personal data | Who can see it | Where | Since | Allowed by rule |
|---|---|---|---|---|
| Customer contact details | Support team14 people | Azure SQL | 2 Mar 2026 | support_read |
| Payment details | Finance team6 people | Databricks | 15 Jan 2026 | finance_read |
| Payment details | Billing exportautomated account | Azure Blob Storage | 20 May 2026 | billing_export |
| Employee records | HR team4 people | PostgreSQL | 3 Nov 2025 | hr_read |
| Health data | Occupational health2 people | PostgreSQL | 11 Feb 2026 | absence_restricted |
You decide what counts as personal data. Sluyz shows who can reach it.
An example of Sluyz Register. Every line names the data, the people, the place, the start date and the rule that allows it.
The whole picture
Including access people have indirectly.
Someone can reach data without being named anywhere: through a group they belong to, or because access to a folder also covers everything inside it. Sluyz works that out, so the overview shows who can really get in.
- Through a group
- Anna is in the Finance team. The Finance team may read invoices. So Anna can.
- Passed down
- Access to a database also gives access to the tables in it, unless something says otherwise.
A data breach
72 hours to report. Know who had access in minutes.
Most of the 72 hours usually goes to finding out who could reach the data. With Sluyz that answer is already there.
Without Sluyz
With Sluyz
Illustration, not a measurement.
What the GDPR asks, and where Sluyz helps.
| Article | What it asks | How Sluyz helps |
|---|---|---|
| Art. 5(2)Accountability | Be able to show that you follow the rules. | Sluyz Register is a dated overview of who has access, and under which rule. |
| Art. 25Data protection by design | Limit access to what is needed. | Access exists only where a rule allows it. |
| Art. 30Records of processing | Keep records, including who receives the data. | The access side of your records stays current without manual work. |
| Art. 32Security of processing | Take appropriate measures and test them. | Access nobody approved is found and removed. Every change is logged. |
| Art. 33Breach notification | Report a breach within 72 hours. | See within minutes who could reach the affected data. |
| Art. 35Impact assessment (DPIA) | Assess the risks of high-risk processing. | The assessment starts from actual access, not from assumptions. |
| NIS2, Art. 21 | Access control policies and incident handling for essential and important organisations. | Every change to access has a proposal, an approval, a time and a name. |
| DORA, Art. 9 | Policies that limit access to data and systems, for financial organisations. | Every change to access has a proposal, an approval, a time and a name. |
Sluyz helps you demonstrate these points. It does not replace your own legal assessment.
Questions legal teams ask us.
Does this make us compliant?
No tool does. Sluyz helps you demonstrate who has access and why, and gives you the evidence. The assessment stays yours.
Who decides what is personal data?
You do. Your organisation marks which data is personal data and in which category. The overview is built from that.
Does it create or change accounts?
Sluyz never creates users or groups. It works with the accounts that already exist in your systems and decides what they may reach.
Can I read the overview without a technician?
Yes. It is a document in plain language with a date. You can export it as a PDF for an auditor or regulator.
Where does it run?
In our cloud in the EU, or inside your own environment.
We are looking for five design partners.
Sluyz works today and has no customers yet. We want to build the next part with a small group of security, legal and data teams.
Ask for a conversationFor security
Know who can reach what. Close it when you must.
Sluyz shows every access to your data across systems and accounts, finds what nobody approved, and gives you one move to shut everything.
Find
Access that nobody approved, found the day it appears.
People leave, suppliers change, someone grants a quick exception. Sluyz compares what exists with what your rules allow, and lists the difference.
- Every system and every accountData platforms and the accounts behind them, in one picture. People, groups and automated accounts.
- What someone can really reachNot only what is granted directly, but also what comes through a group or is passed down from a parent, such as a database to its tables.
- Evidence for NIS2 and DORAEvery change has a proposal, an approval, a time and a name. It helps you demonstrate access control and incident handling.
| Who | Can reach | Why it is flagged |
|---|---|---|
| Left the company | Databricks | Account still active after leaving |
| Left the company | PostgreSQL | Account still active after leaving |
| Old supplier login | Databricks | Contract ended, login never removed |
| Old supplier login | Azure SQL | Contract ended, login never removed |
Each line can be removed with one approved change.
Stop
When the storm comes, one barrier closes every gate.
During a hack, Storm Barrier closes all access to all systems in one move, accounts included. Afterwards, the last approved situation is put back the same way.
More about Storm Barrier- DatabricksopenclosedClosed
- Azure Blob StorageopenclosedClosed
- Azure SQLopenclosedClosed
- PostgreSQLopenclosedClosed
- Microsoft Entra IDopenclosedClosed
- Incident teamstays openClosed
Try it: let both people agree, then close everything. This is a demonstration, not connected to a real system.
We are looking for five design partners.
Sluyz works today and has no customers yet. We want to build the next part with a small group of security, legal and data teams.
Ask for a conversationStorm Barrier
When the storm comes, one barrier closes every gate.
During a hack, Storm Barrier closes all access to all systems in one move, accounts included. Afterwards, the last approved situation is put back the same way.
Try it
Two people agree. Then everything closes.
- Two people must agreeFor example the CISO and the incident manager.
- Your incident team keeps accessAccounts you choose in advance stay open.
- Everything is recordedEvery step is logged as evidence for the regulator.
- DatabricksopenclosedClosed
- Azure Blob StorageopenclosedClosed
- Azure SQLopenclosedClosed
- PostgreSQLopenclosedClosed
- Microsoft Entra IDopenclosedClosed
- Incident teamstays openClosed
Try it: let both people agree, then close everything. This is a demonstration, not connected to a real system.
What happens, in order.
- 1
Something is wrong
Your team sees an attack, or suspects one.
- 2
Two people agree
For example the CISO and the incident manager. One person alone cannot close everything.
- 3
Everything closes
Access to data and the accounts behind it is withdrawn in one move. Your incident team keeps its own way in.
- 4
The approved situation returns
When it is safe, the last approved situation is put back. You see the plan first.
Close as much as you need, no more.
Not every incident needs everything shut. You choose how far Storm Barrier goes.
- Everything
- All systems and all accounts.
- One system
- For example only Databricks.
- One set of data
- For example only payment details.
- One account
- For example a login you think is stolen.
You can also rehearse: a dry run shows what would close without closing anything.
We are looking for five design partners.
Sluyz works today and has no customers yet. We want to build the next part with a small group of security, legal and data teams.
Ask for a conversationFor technical teams
One rule instead of four dialects.
The same access rule has to be written differently for every platform. With Sluyz you write it once in UAL and review the change before it is applied.
GRANT SELECT ON TABLE prod.finance.invoices TO `finance-analysts`;
GRANT SELECT ON OBJECT::finance.invoices TO [finance-analysts];
GRANT SELECT ON finance.invoices TO finance_analysts;
az role assignment create --role "Storage Blob Data Reader" --assignee-object-id <group> --scope <container>
# tagged: payment_details
policy finance_read {
allow read
on finance.invoices
to group finance-analysts
}
The UAL shown here is an illustration of the idea, not final syntax.
Three layers, from what each platform stores to who can really get in.
Sluyz keeps the platform’s own data, its own stable model, and the answer to “who can reach what” apart. Each layer is built from the one below.
Effective access
Explains what each identity can actually reach, including access through group membership and through object inheritance.
Feeds Sluyz Register
Annais inFinance teammay readfinancecontainsinvoices
So Anna can read invoices, although her name is in no grant.
The stable model
One model of principals, objects and grants that reads the same for every platform. UAL compiles to this, and changes are planned against it.
Where plans are made
- Principal
- group finance-team
- Object
- schema finance
- Grant
- read
Provider-native data
Grants and membership exactly as each platform and directory stores them.
Read from each source
- Entra ID
anna is a member of finance-team- Databricks
GRANT SELECT ON SCHEMA finance TO `finance-team`
UAL never creates users or groups.
The principals in a grant refer to identities that already exist in a platform or in your directory. Who exists is decided where it is decided today. UAL decides what they may reach.
How a change travels.
- 1
Describe
Write intended access once in UAL.
- 2
Compile
Sluyz turns it into the stable model (L2), free of platform dialect.
- 3
Compare
It reads the provider-native state (L1) and shows the difference.
- 4
Plan
You review exactly what will be added or removed.
- 5
Apply
After approval, changes are made with each platform’s own controls.
What it works with.
A directory is a source like any other. That is why Storm Barrier reaches accounts as well as data.
- Rules are written in UAL, an open language. A free command-line tool works with one platform.
- Changes are made with each platform’s own access controls. Nothing sits between your users and your data.
- Runs in our EU cloud or inside your own environment.
| System | Kind | Status |
|---|---|---|
| Databricks | Data platform | Works today |
| Azure Blob Storage | File storage | Works today |
| Azure SQL | Database | Works today |
| PostgreSQL | Database | Works today |
| Microsoft Entra ID | Directory: accounts and groups | Works today |
| Snowflake | Data platform | Coming |
| Amazon Redshift | Data platform | Coming |
| Salesforce | CRM | Coming |
| Microsoft Dynamics | CRM and ERP | Coming |
We are looking for five design partners.
Sluyz works today and has no customers yet. We want to build the next part with a small group of security, legal and data teams.
Ask for a conversationDesign partners
We are looking for five design partners.
Sluyz works today and has no customers yet. We want to build the next part with a small group of security, legal and data teams.
What you get.
- A free check of one system: who can reach what, and what nobody approved.
- A say in which systems we add next.
- Partner pricing when you decide to stay.