Kryoverse
Start a project

Finance · AI development

ChatGPT-style help that keeps customer data private.

A finance company's customers ask questions and get answers from the company's own material. The whole system runs inside the client's environment.

A private GPT inside their own environmentA locked boundary with a question and an answer inside, encryption hatching on the border.QAinside their environment

The problem

A finance company wanted to offer its customers ChatGPT-style help. Customers would ask in their own words and get a clear answer from its material.

A public chatbot was out of the question. Customer data could not end up in someone else's logs. It also had to meet GDPR and the security rules the company is audited on.

What we built

  • Customer chat assistant

    A ChatGPT-style chat where customers type a question and get a clear answer back.

  • Answers from company material

    Replies come from the company's own documents, not the open internet.

  • Runs in their environment

    The whole system runs inside the client's own environment. Customer conversations stay with them.

  • Encryption and access control

    Conversations are encrypted. Only the people and systems that need access get it.

  • GDPR data handling

    Customers stay in control of their data. It is stored and deleted the way GDPR requires.

How AI helps

  • Searches first, then answers

    The AI finds the right parts of the company's material, then answers from them.

  • Improves within the rules

    It gets better from real conversations, but only in ways the privacy rules allow.

  • Reviewed before launch

    Before customers saw it, the client checked its answers against their own documents.

How it works

  1. 01

    A customer types a question into the assistant.

  2. 02

    The assistant searches the company's own material for the answer.

  3. 03

    It writes a clear reply based on what it found.

  4. 04

    The conversation is encrypted and stored under the company's privacy rules.

  5. 05

    Every step happens inside the client's environment.

Built with

Frontend
ReactTypeScript
Backend
PythonFastAPI
AI
LangChainRAGVector search
Data
PostgreSQL
Infrastructure
DockerKubernetes

What changed

  • Customers get clear answers from material the company trusts.
  • Customer data stays inside the company's own environment.
  • Security was checked at every stage, so the team can explain it to an auditor.

Next case study

Distribution and logistics · AI development

An AI agent for invoices and orders

Reads invoices, orders, and delivery notes from email. Checks them against the ERP and flags what does not match.

Read the case study
See all work

Need something like this? Let's build yours.

Start a project