[nonimo]
EN
Download the app

Nonimo Gateway for UK teams

Say yes to AI, and keep every bit of personal data inside your network.

Nonimo Gateway pseudonymises personal data on your own servers before each AI call and puts it back in the reply. It sits in a container in front of the OpenAI API, and your services need no code changes beyond the base_url.

  • Runs on your own servers
  • Your content does not touch Nonimo
  • NI numbers, NHS numbers and bank details
POST /v1/chat/completions
  1. Sent from your staff chatbot

    Summarise the case notes for Hannah Wright and email her at hannah.wright@example.com.

  2. What the provider gets masked on your network

    Summarise the case notes for [PERSON_1] and email her at [EMAIL_1].

  3. The provider replies

    Dear [PERSON_1], we have reviewed your case and everything is in order.

  4. Back to your service real details

    Dear Hannah Wright, we have reviewed your case and everything is in order.

Fictitious data. Anything that calls AI through the API passes through the Gateway; ChatGPT in a browser does not (for that, see Nonimo for organisations).

Getting started

PII redaction for your OpenAI API integrations, set up in an afternoon

Your services’ code stays untouched: one container on your server and a new base_url.

  1. A container in your data centre or your cloud.

    Technical detail

    It installs on your servers or in your cloud subscription, alongside your services, and you can run several instances alongside each other.

  2. Your code carries on speaking the OpenAI API. Only the address changes.

    Technical detail

    The Gateway speaks the OpenAI API. Point your services at it and they carry on calling /v1/chat/completions or /v1/responses, with function calling. Upstream it can be OpenAI or any compatible API.

  3. Personal data leaves as tokens and returns as the real details in the reply.

    Technical detail

    Every request is checked in full before it goes to your provider: personal data is replaced with tokens and, in the reply, each token is swapped back for its real value. Your services’ AI requests pass through it before they leave your network.

The official OpenAI SDK works without changes; simply point the client at the Gateway:

Python
from openai import OpenAI

# Same client, same key: only base_url changes
client = OpenAI(base_url="http://your-server:8080/v1")
resp = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Summarise the case notes for Hannah Wright."}],
)
Node
import OpenAI from "openai";

// Same client, same key: only baseURL changes
const client = new OpenAI({ baseURL: "http://your-server:8080/v1" });
const resp = await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Summarise the case notes for Hannah Wright." }],
});
curl
# Your usual request, sent to your own server
curl http://your-server:8080/v1/chat/completions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [{"role": "user", "content": "Summarise the case notes for Hannah Wright."}]
  }'

Restored on the way back

Pseudonymisation that gives the real details back

Your provider only ever sees tokens; your service receives the reply with the real details, each one back in its place.

servicedesk.example.com

Staff chatbot

Summarise the ticket Daniel Hughes[PERSON_1] raised from daniel.hughes@example.com[EMAIL_1].

Reply

The ticket from Daniel Hughes[PERSON_1] concerns a VPN log-in fault since Monday. I am emailing daniel.hughes@example.com[EMAIL_1] the steps to reconnect.

Tokens restored: 2 of 2 Personal data sent: 0

Customer service

Draft an email to Charlotte Evans[PERSON_1] (charlotte.evans@example.com[EMAIL_1]) confirming that we have received her paperwork.

Reply

Dear Charlotte Evans[PERSON_1], thank you: we can confirm we have received your paperwork. We will write to charlotte.evans@example.com[EMAIL_1] as soon as everything is ready.

Tokens restored: 2 of 2 Personal data sent: 0

Agent with tools

Check the referral status for Amelia Patel[PERSON_1], NHS number 123 456 7890[NHS_NUMBER_1].

The model calls a function

find_referral({ "patient": "Amelia Patel[PERSON_1]", "nhs_number": "123 456 7890[NHS_NUMBER_1]" })

Your tool receives the real value; the provider only ever saw the token.

Tokens restored: 2 of 2 Personal data sent: 0

HR

Prepare the new-starter letter for George Clarke[PERSON_1], National Insurance number QQ 12 34 56 C[NINO_1].

Reply

Dear George Clarke[PERSON_1], welcome to the team. We have recorded your National Insurance number, QQ 12 34 56 C[NINO_1], for payroll.

Tokens restored: 2 of 2 Personal data sent: 0
  • Every detail, its own tokenIt does not rewrite or summarise, and each detail keeps its token throughout the conversation.
  • Detection runs within your networkYour data does not leave your network to be checked.
  • Function callingYour tools receive the JSON arguments with the real values.

For your DPO: this is a pseudonymisation measure within the meaning of Article 4(5) UK GDPR, and the mapping back to the real details stays in your infrastructure. What the ICO says, in the questions below.

We explain how pseudonymisation differs from anonymisation in pseudonymisation under the UK GDPR.

Technical detail

Detection runs within your network: your data does not leave it to be checked. It does not rewrite or summarise, it only marks what is personal data, and within a conversation each detail keeps the same token from beginning to end. In tool calls, the provider sees tokens and your tool receives the real values.

What goes out to OpenAI

Draft an email to [PERSON_1] ([EMAIL_1]) confirming that we have received her paperwork.

What your service receives

Dear Charlotte Evans, thank you: we can confirm we have received your paperwork. We will write to charlotte.evans@example.com as soon as everything is ready.

UK identifiers and more

PII redaction for National Insurance numbers, NHS numbers and names

Every document, in its own format. Choose a country and see what reaches the AI.

United Kingdom: also recognised

  • NameSophie Turner[PERSON_1]
  • Mobile+44 7700 900461[PHONE_1]
  • Sort code and account00-00-00 00000000[BANK_ACCOUNT_GB_1]
  • PostcodeAB1 2CD[POSTCODE_1]
  • Emailsophie@example.com[EMAIL_1]

And in every country

  • API keys[API_KEY_1]
  • Bank cards[CARD_1]
  • Passports (MRZ)[PASSPORT_1]
your service → Gateway → api.openai.com
  • NI number[NINO_1]
  • NHS number[NHS_NUMBER_1]
  • UTR[UTR_1]
  • Driving licence[DRIVING_LICENCE_1]

Your service sends

Update the payroll record for Oliver Bennett, NI number QQ 98 76 54 A.

The AI receives

Update the payroll record for [PERSON_1], NI number [NINO_1].

Invented details and sample numbers that are not valid.

See the full table of identifiers by country

The Gateway recognises the everyday personal data (names, email addresses, phone numbers, postal addresses, API keys) and, in addition, each country’s identifiers in their own format. For the UK that means National Insurance numbers, NHS numbers, Unique Taxpayer References and DVLA driving licence numbers. It also covers IBANs, bank cards and passports (their machine-readable zone).

CountryIdentifiers
United KingdomNational Insurance number, NHS number, Unique Taxpayer Reference (UTR), DVLA driving licence number
IrelandPPS number
United StatesSocial Security number
AustraliaTFN, Medicare
GermanySteuer-ID, Sozialversicherungsnummer
FranceNIR (numéro de sécurité sociale), SIREN, SIRET
SpainDNI, NIE, NIF, Social Security number
Italycodice fiscale, partita IVA, carta d’identità elettronica (CIE), tessera sanitaria
PortugalNIF, NISS
BrazilCPF, CNPJ, CNH (driving licence)
All countriesIBAN, passports (machine-readable zone), bank cards

Compared with Presidio

Weighing up Presidio? Nonimo Gateway arrives ready to deploy

Presidio finds personal data in text. Putting it in front of the OpenAI API means building a compatible proxy, and that is down to your team. Nonimo Gateway comes with one.

Down to your team: 6 of 8 tasks Down to your team: point base_url at the Gateway
What each option provides and what your team has to build
Task Building it on Presidio With Nonimo Gateway
Find personal data in text Presidio provides it Included
OpenAI-compatible proxy You build it Comes ready
Place it in front of your AI provider You connect it Repoint base_url
Restore the real details in the reply You build it Comes ready
National identifiers by country Some built in (incl. NHS, NINO); you choose and tune the rest Those in the table above
Text in languages other than English You configure it Comes ready
Keeping it up to date Your team We supply each new version
Where it runs On your infrastructure A container on your server

The full comparison, row by row: Microsoft Presidio alternative →

Technical detail

Presidio finds personal data in text, but it does not include an OpenAI-compatible proxy: to place it in front of the API, your team has to build that proxy, connect it and look after it.

Nonimo Gateway provides that part ready-made: a container on your own server, an OpenAI-compatible API, the real details restored in the reply and the national identifiers in the table above.

Your infrastructure

Under your control, on your own network

You keep your AI provider, and the Gateway becomes one more component inside your perimeter.

LiteLLM

Running LiteLLM already? Switch Nonimo on as a guardrail

If your platform already routes its calls through LiteLLM, Nonimo fits in in one of two ways. Choose the one you need.

Guardrail in LiteLLM

  1. Your services
  2. LiteLLM
  3. Nonimo
  4. The model

Out Summarise the case notes for [PERSON_1]

Back The reply arrives with the tokens: [PERSON_1]

  • Personal data is masked before it reaches the model
  • Switched on in the LiteLLM configuration
  • The real details are not restored in the reply
Need them back? Use proxy mode →

Gateway in proxy mode

  1. Your services
  2. Nonimo Gateway
  3. Your AI provider

Out Summarise the case notes for [PERSON_1]

Back The reply arrives with the real details: Hannah Wright

  • Personal data is masked before it reaches the model
  • The reply returns with the real details
Technical detail

Running LiteLLM already? Nonimo switches on as a guardrail, or as a proxy in front of your provider if you want the real details back in the reply.

Plans

Choose the PII redaction plan that fits your organisation

There’s a 30-day trial on every plan. Should you exceed your monthly word allowance, the Gateway alerts you and your service carries on uninterrupted.

600 staff

Right for you Nonimo Gateway Business, £419/month

Practice

The place to start

For practices, chambers and smaller teams

Nonimo Gateway Practice £249/month Nonimo Gateway Practice or £2,490 a year

Up to 25 million words a month Up to roughly 200 staff
  • 1 instance
  • Email support
Request access

Business

For larger organisations running several services

Nonimo Gateway Business £419/month Nonimo Gateway Business or £4,190 a year

Up to 150 million words a month Up to roughly 1,000 staff
  • Up to 3 instances
  • Offline mode on request
  • Priority support
Request access

Enterprise

For groups with their own requirements and terms

Custom pricing

  • No usage cap
  • Unlimited instances
  • Offline mode on request
  • Your own contract and terms
Get in touch
How many staff is each plan designed for?

By our estimate, the Practice plan’s 25 million words a month will serve up to roughly 200 staff. Business, at 150 million, stretches to roughly 1,000. Enterprise is uncapped and intended for organisations above 1,000 staff or spread across several sites. If your usage exceeds the allowance, we’ll tell you and nothing stops working.

Access

Request access to Nonimo Gateway

Tell us what your stack looks like today and we will arrange a trial on your own server, with your own security team in the loop.

Which AI services does your team use at the moment? (tick any that apply)

We only use these details to get back to you about Nonimo Gateway. Read our privacy notice.

Questions

Questions from IT and data protection teams

What heads of IT, security leads and DPOs ask before they deploy the Gateway.

What it does and what it covers

What exactly does the AI provider see?

The prompt with tokens such as [PERSON_1] or [NINO_1] in place of the personal details, sent to your provider. If your policy blocks a data type, the request is stopped before it leaves and the provider receives nothing. The token map stays on your server.

Does detection rely on an external service?

No. Detection runs locally, within your network.

Which APIs and features are supported?

The OpenAI API and any API compatible with it: Chat Completions and Responses, with tool or function calling.

How many staff is each plan designed for?

By our estimate, the Practice plan’s 25 million words a month will serve up to roughly 200 staff. Business, at 150 million, stretches to roughly 1,000. Enterprise is uncapped and intended for organisations above 1,000 staff or spread across several sites. If your usage exceeds the allowance, we’ll tell you and nothing stops working.

Security and data protection

Is this pseudonymisation or anonymisation under the UK GDPR?

Pseudonymisation, as defined in Article 4(5) of the UK GDPR: the data can no longer be attributed to a specific individual without additional information, which is kept separately, and here that information stays in your infrastructure. It is not anonymisation, although many people search for ‘data anonymisation’ when this is what they need. It helps with data minimisation, but the ICO treats pseudonymised data as personal data, so your lawful basis, the Article 28 terms with your provider and any DPIA still apply.

What network access does it need?

Access to your AI provider, plus a connection to Nonimo for the licence and usage figures, not your content. If your network cannot reach Nonimo, the Business and Enterprise plans offer an offline mode on request.

What happens if the Gateway or the provider fails?

By default, the request is not sent and your application gets a clear error.

Does the Gateway connect to Nonimo?

Only for the licence and usage figures, not your content. If your network cannot reach Nonimo, the Business and Enterprise plans offer an offline mode on request. Write to us at hello@nonimo.ai and we will agree the terms.

Deployment and licensing

How is it deployed?

A server or a cluster that runs containers, in your data centre or your cloud. It installs alongside your services, and you can run several instances.

How is it licensed?

With a signed licence for each server or instance, renewed automatically on connected plans and loaded by hand when running offline, on terms agreed with each customer. If you go over your plan’s usage, the Gateway warns you and your service carries on.

How are updates handled?

Your IT team decides when to apply updates. The Gateway does not update itself, and it lets you know when a new version is available.

Guides

Further reading for your IT and security team

Guides on AI gateways, privacy proxies and keeping personal data out of your LLM API calls.

Get started

Try PII redaction on your own server before your next AI release

Request access, put the Gateway in front of your OpenAI integrations and let your security team test it on your own estate. For the AI tools staff open in a browser, see Nonimo for organisations.

Tools

Remove personal data from a file or text before AI sees it. Or cover the details on a copy of your ID and add a watermark before you send it.

All tools →