Kronstadtsoft

AI consulting and business integration

We offer consulting and build AI integrations into your company’s processes and systems. We build agents and orchestration, MCP servers, the platform your teams use to reach the models, and search over internal documents. Every solution takes the GDPR and the EU AI Act into account.

Diagram · The company AI platform
TEAMS AND AGENTSASSISTANTS · WORKFLOWSAI PLATFORMROUTING · BUDGET · LOGMODELSZERO DATA RETENTIONDOCUMENTSSEARCH WITH SOURCESCOMPANY SYSTEMSERP · CRM · VIA MCPTEAMS AND AGENTSASSISTANTS · FLOWSAI PLATFORMROUTING · LOGMODELSZERO DATA RETENTIONDOCUMENTSSEARCH WITH SOURCESCOMPANY SYSTEMSERP · CRM · VIA MCP

Scope

I

What we integrate.

  1. 01

    AI consulting

    Strategy · Architecture

    We help you make the technical decisions before you invest: which processes are worth automating, which architecture to use, which providers and models to choose for each task, and which GDPR and EU AI Act obligations apply. We work at €60 per hour, on your systems and your data, and you receive the recommendations in writing.

  2. 02

    Assessment and integration plan

    Processes · Plan

    We review your processes and the data they use, classify the data by sensitivity and place each AI system in its risk category under the EU AI Act. We deliver a written plan: the proposed architecture, the integrations it needs, the estimated monthly cost and the criteria we measure the result against.

  3. 03

    Agents and orchestration

    Agents · Workflows

    We build agents and automated workflows that work directly inside the systems you already run: ERP, CRM, email, e-invoicing, documents. An orchestration layer splits the work between several models, classifiers and tools. Important decisions go through an approval step by a person. Every action stays in the log.

  4. 04

    MCP servers and integrations

    MCP · API

    We connect your company’s systems to the AI assistants your team uses, such as Claude, ChatGPT or Copilot, through the Model Context Protocol. The MCP servers have OAuth sign-in, per-user permissions and an audit log, and the assistants see and do only what you allow.

  5. 05

    Company AI platform

    Gateway · Models

    Every call to a model goes through a single gateway. The gateway gives routing between providers, budgets per team, logging, removal of personal data before it is sent, and providers with zero data retention. The orchestration, the classifiers and the document search run on this platform. For each task, we choose the models by testing them on your data. This is also where we set up the working environment for agents and coding assistants.

  6. 06

    Search over internal documents

    RAG · Documents

    We build semantic search over your company’s documents: your team asks questions and gets answers with a reference to the source. Each user sees only the documents they already have access to.

  7. 07

    Quality, security and operation

    Tests · Monitoring

    We check every system against test sets before launch and after every change of models, instructions or data. We test resistance to instructions hidden in documents and emails, track quality and cost in production, and operate the system every month.

  8. 08

    Compliance and training

    GDPR · AI Act

    We prepare the documents the GDPR requires for AI workflows: the data processing agreements with the providers, the record of processing and, where needed, the impact assessment. We keep the inventory of AI systems, apply the transparency obligations of the EU AI Act and prepare the training program for the staff who use AI, as Article 4 requires.

Data protection

II

How we protect the data.

Principles

Zero data retention
we use model providers that, by contract, do not keep prompts and responses or use them for training; we confirm the providers in writing for each project
Hosting
the application runs in the European Union or in your company’s own cloud account; the plan states where the models run
Access
role based, with an audit log for every action
Personal data
removed or masked before it reaches the models, where it is not needed
Retention
only the data you decide to keep, for the period you set, then deleted automatically

Kronstadtsoft has signed the European Commission’s AI Pact pledges and is on the public list of signatories. See the AI Pact signatories

Work

III

What we have already built.

  1. 01

    Patrimoniu

    Own product

    A portfolio tracking application built around AI agents: through its own MCP server, your AI assistant reads and manages the portfolio, within the limits you approve.

    See the project

How we work

IV

From the first conversation to operation.

  1. 01

    First conversation

    Free, one hour

    We talk through the processes you want to improve and the data they use. Together we decide where AI helps and where to start.

  2. 02

    Assessment and plan

    Offer

    We send an offer for the assessment, sized to your company. At the end you receive the written plan, with the architecture, the estimated monthly cost and the acceptance criteria.

  3. 03

    Development

    Fixed price

    We build the integration at the fixed price in the written offer. The acceptance criteria are agreed in writing before the work starts.

  4. 04

    Operation

    Monthly

    We monitor quality and cost, update the models and send you a monthly report of what changed.

Rates

V

What it costs.

Terms

Consulting
€60 per hour
First conversation
free, one hour, once per company
Assessment and plan
an offer for each company
Operation
a monthly offer
Model usage
billed by usage, estimated in the plan

Book the first conversation

Tell us what you want to build or improve. We propose a few days and times for the first conversation.

Send us a messagecontact@kronstadtsoft.ro