Self-hosted data protection for AI
Your teams use AI. Protect the data before AI sees it.
Salus replaces sensitive values with persistent, category-typed tokens before content reaches the AI. Your teams keep their approved tools; authorised answers are restored inside your environment. Fast, deterministic text detection runs on CPU.
31+ detection classespurpose-built modelsCPU text detectionpersistent chat tokensOCR on requestsemantic policies on requestOpenShiftSIEM events
Müşteri Ayşe Demir (TCKN 12345678901) mart faturasına itiraz ediyor…
[PERSON_81af3c]TR12 3456 …→[IBAN_c412e8]+2 moreSelf-hosted
Runs in your datacentre or private cloud. Salus the company holds no copy of your data or your keys.
Multilingual
Small specialised models and validated rules work over Turkish, English and other-language documents. Language and identifier coverage are validated on your files.
Any provider
OpenAI, Anthropic, Google, Mistral or your own model. You keep your contract and your credentials.
Two-way
Answers come back with the real values restored, for the people authorised to see them.
The boundary
Only tokens cross.
Protected values become category-typed tokens before content reaches the model. Authorised results are restored on your side.
Names and identifiers on the left become typed tokens as they cross the boundary; the answer comes back and the tokens become names again, inside your network.
Why tokens
Blocking stops the work. Redaction ruins the answer.
Most tools that protect AI traffic either refuse the prompt or black out the sensitive parts. Both keep the data safe. Neither leaves your colleague with an answer they can use, so people go around them.
“Maria Sandoval missed three payments. Her guarantor Erik Lindqvist missed none. Draft a reminder letter to the right person.”
Block
The model receives: nothing. The prompt is stopped.
Your colleague gets: “This request violates company policy.”
Safe, and the work moves to a personal account.
Redact
The model receives: “▮▮▮ missed three payments. Her guarantor ▮▮▮ missed none.”
Your colleague gets: “Dear ▮▮▮, our records show three missed payments.”
No name, and no way to tell which of the two it is for. Safe, and the answer has to be rebuilt by hand.
Tokenise
The model receives: “[PERSON_81af3c] missed three payments. Her guarantor [PERSON_5b90da] missed none.”
Your colleague gets: “Dear Maria Sandoval, our records show three missed payments.”
Addressed to the right person, ready to send. Safe, and the answer is usable as it arrives.
Proof
Measured on real business documents, not a demo set.
Figures from the Salus Benchmark Report, October 2026, scored on documents held out from model training. Your acceptance test repeats the measurement on your own documents and policy.
- 3,396real business documents
- Held out from model training. 6,754 pages, 94,127 labelled values.
- 98.3%personal-data values fully protected
- 97.81% across all 94,127 values, organisations included.
- 3.8%over-masking
- Non-sensitive text masked on the same run. 96.2% masking precision.
- 0.36 smedian per document on CPU
- The full batch in 14.5 minutes on one 32-vCPU server. No GPU for text detection.
Salus Benchmark Report, 26 October 2026. Nine systems measured on the same workload.
How it works
Four steps. One of them leaves your network, and it carries tokens.
Detect
Small specialised models and validated rules find sensitive values. Optional OCR reads scanned pages locally before the extracted text is protected.
Tokenise
Each value becomes a typed token such as [PERSON_81af3c]. The same person keeps the same token for the whole case, so the model can still follow who did what.
Your AI call
Your application calls its own provider with its own credentials. Salus does not call the model for you. Only tokens cross the boundary.
Restore
When the answer returns, Salus puts the real values back inside your network, for users who are authorised to see them.
What it detects
Grouped by what a reviewer would ask about.
Core classes cover names, contact details and structured identifiers. International and sector formats are delivered for your jurisdictions; your own identifiers are scoped per deployment. OCR and finance or telecom semantic policies are available on request.
Full detection dictionaryIdentity and organisation8
- Place of birth
- Nationality
- Person name
- Organisation
- TCKN11 digits, two-step checksum
- Passport number
- Date of birth
- Vehicle plateTurkish plate grammar
Contact and location7
- Phone number (MSISDN)Numbering-plan check, E.164 and national formats
- E-mail address
- Postal address
- Location
- Coordinates
- Identifying URL
- IP addressIPv4 and IPv6 syntax
Financial and government identifiers8
- MERSIS number
- UETS address
- Social-security number
- IBANCountry length and structure, mod-97 check
- Payment card numberLuhn check, 13 to 19 digits
- Tax ID10 digits, checksum
- Bank account number
- Customer or subscriber number
Telecom and device3
- IMSI15 digits, MCC and MNC structure
- ICCID19 or 20 digits, Luhn check
- IMEI15 digits, Luhn check
Credentials, international and sector identifiers7
- Foreign identity, tax and social-security numbers
- Driving licence numbers
- MAC addresses
- Credentials
- Domestic banking formats
- Healthcare identifiers
- Crypto-asset addresses
Signatures, identity documents and faces3
- Handwritten signatures
- Identity documents
- Faces
Customer-defined1
- Your own identifiers
Semantic layer 2
Optional and complementary. Masks a stated disclosure about an identifiable person; generic statements pass unchanged. It only adds protection; it never removes or re-types what the core found.
- Customer financial data
- Telecom usage
Where it fits
Two ways to put Salus in the path.
Integrate it into the applications you build, or deploy it around the ones you don't control. Both run on the same engine, with the same tokens and the same rules.
In the applications you build
Call Salus from your code, your AI gateway or your batch jobs: protect before the provider call, restore after it. MCP tool-result flows can use a client or server wrapper around the Engine API, scoped for the deployment.
- In-house assistants, chat apps and RAG
- LLM gateways, including the LiteLLM guardrail
- MCP tool-result integration through the Engine API
- Document processing and batch analytics
- Agents and automated pipelines
Around the applications you don't control
For the AI tools your teams use directly, Salus is deployed on managed endpoints and protects what is typed or attached, without changing the application.
- Browser-based AI services
- Desktop AI applications
- Documents and attachments
- Optional human review before anything is sent
Security and compliance
Reviewed like any other system inside your perimeter.
Because that is where all of it lives. Controls are named and numbered, frameworks are mapped, and nothing is claimed as certified.
Controls
- Named control catalogue, SC-01 onward
- One decision record per request
- Stable reason codes
Published with the datasheet.
Frameworks · mapped
- OWASP Top 10 for LLMs
- NIST AI RMF
- ISO 27001 Annex A
- KVKK
- BTK traffic data categories
Mapped, with evidence aligned. Not certified.
Deployment
- On-premises
- Private cloud
- Isolated network with a mirrored registry
Containers, CPU, PostgreSQL state.
Data
- No Salus-side copy
- No keys leave your environment
- No telemetry home
Nothing on our side to breach or subpoena.
Control
You run it. You hold the keys.
Runs where you decide
On-premises, private cloud or an isolated environment. Text detection runs on CPU; scans and the optional Semantic Layer use GPU acceleration. State stays in your PostgreSQL.
No Salus-side copy
Detection, tokenisation and the mapping back to real values all stay in your environment.
Fails closed
If detection or tokenisation cannot complete, the request is stopped. Nothing goes out unprotected because a step was skipped.
Audit without the data
Every protect and restore is logged for your SIEM. The logs record what happened without reprinting the values you were protecting.
Administration and operations
Operate it like any other system in your perimeter.
Configure administration, identity integration, workload policy and review retention for your deployment. Keep state, keys and operational evidence inside your environment.
Administration console
Console for organisations, keys, permissions, rate limits, policy and detection review. Languages and administrative integrations are configured per deployment.
Your directory, your roles
Active Directory, LDAP or single sign-on can be configured for the deployment. Map your groups to roles and separate administration, review and restoration permissions.
Policy per workload
Agree the categories, handling actions and identifier formats for each workflow. Organisation-scoped API keys and permissions control calling systems.
Security and review evidence
Security events carry stable codes without sensitive values. Review runs are recorded in your PostgreSQL on a best-effort path, with loss alerts; values are encrypted with your key and retained under your policy.
Monitoring you already have
OpenMetrics on every service. Dashboards and alerts are delivered in your monitoring platform, with a documented CEF mapping for your security collector.
Compute sized for the workload
Core text detection runs on CPU. Scanned-page OCR and the optional Semantic Layer use GPU acceleration. Capacity and availability are sized for the agreed deployment.
Models and rules are versioned artefacts pinned to each release. A release is either active or rolled back; nothing retrains silently.
Company
It started with a problem we kept seeing.
AI note takers in every meeting, client files pasted into chat assistants, and organisations that could not adopt any of it because of what the data contained.
An engineer and a lawyer, we saw the same picture from both sides. Teams were already using the best models on real work with no protection at all, because every “safe” alternative was too weak to be worth using. Regulated organisations either looked away or stayed out. Salus was built to remove that choice: the best models available, working on real files, without ever learning who the files are about.
The same problem exists wherever data is sensitive and regulated: banks, telecom operators, insurers, law firms. Salus is built for them, in every language their teams work in. In test with large regulated organisations in Türkiye.
- Language model training and evaluation
- Rust systems and enterprise platforms
- Telecom and banking delivery
- KVKK, contracts and compliance
- Regulated-sector operations
Meet the founders
Software engineer since 2015 and the architect of the detection stack: trains and evaluates the models, owns the benchmarks and the release gates. Previously built AI agent products at Fundamental Research Labs, a San Francisco AI company. Founded Paramosoft in 2020, consulting on cybersecurity and fraud compliance for a participation bank’s digital channels.
Fifteen years building and running enterprise platforms. Wrote the Rust services behind detection, verification, redaction and restore. Before Salus: CRM modernisation and e-SIM provisioning at a national mobile operator, CPaaS microservices at Netaş, technical lead of a video-marketing SaaS. M.Sc. Computer Science.
Salus is a product of Paramosoft Yazılım Hizmetleri Ltd. Şti., Istanbul.
Pilot
Start with one team, one workflow, your own data.
A focused pilot is the fastest way to see what Salus catches in your traffic, and what it hands to the model instead. Tell us the workflow and we will set it up with you.
[email protected]