Skip to content

STRATEGICRepublic of San Marino

AboutContact
ItalianoEnglish
Client area
Contact

AI experts for businesses

Artificial intelligence for businesses, from first analysis to everyday use

Trying out artificial intelligence is easy. Making it work every day inside your ERP, archives and processes is a different job. It takes development, databases and security in the same project. And when AI is not the answer, we tell you before we start.

Let’s talk about your situationOur AI services
Illustration of a process analysis with a traffic light for each department, and an assistant’s answer checked by a person.
Example · the data in this illustration is not real
Development · databases · security
in the same team
On the systems you already have
ERP, databases, archives and email
Handed over in full
code, configuration and instructions stay with your company

01 · copy the link to this sectionThe situation

Why you need a team, not just an AI vendor

A prototype left half-finished costs you twice. You pay for the work already done, and your people keep losing hours.

The trial usually impresses. The problems start with real data. The ERP has no modern interfaces. The archive holds three versions of the same document. Security asks where the data goes, and Legal asks who is accountable for a mistake.

At that point, data, connections and security matter more than the model. That is why development, databases and security work on the same project.

Example of procedures and ERP data connected to an AI assistant, with the answer checked by a person.

Built for everyday use

It copes with real volumes, handles errors and receives scheduled updates.

Data read by database specialists

The assistant reads the ERP without changing it or slowing it down.

Security decided at the start

Permissions and logs are project requirements, checked before go-live.

Handed over in full

Code, configuration and instructions stay with your company.

02 · copy the link to this sectionUse cases

Where an AI project usually starts

AI pays off most where the work is made up of documents, rules and data the company already has, and where a person checks the result.

AI Procedures and manuals in different versions across several sites Every answer shows the document and the current revision. Company Brain AI Orders, specifications and contracts retyped by hand AI extracts the data and prepares it in the ERP. Exceptions go to a person. AI agents and integrations AI An assistant already live and giving poor answers It cites outdated documents or reveals confidential content. It needs fixing, rebuilding or switching off. AI security and governance AI A vendor’s proposal to assess Before you sign, you get an independent review of data, costs and exit terms. Second opinion DBA ERP figures for management Ask about the data in plain language. Nothing can be changed, and the query is shown. Oracle & Database Engineering LAB A system the market doesn’t offer We choose the technology once we understand the problem. Every change goes through review and testing. Bespoke software

03 · copy the link to this sectionWhat we do

Our artificial intelligence services

Use them on their own or one after another.

Example of an answer citing its sources, a procedure, a manual and an ERP table, with permissions by department. AI Company Brain Ask your manuals, procedures and ERP data questions in plain language. Each answer shows the source document, and permissions are set by department. View the service Example of an operation prepared by an AI agent, waiting for a person to confirm it. AI AI agents and integrations AI prepares the work in your ERP, CRM and email. A person confirms the actions that matter. View the service Example of a register of AI systems with a blocked request. AI AI security and governance A few clear rules of use, and security testing of the assistants that read company data. View the service Example of an AI assistant contract with notes in the margin. AI Second opinion on AI projects and vendors Proposals and contracts reviewed before you sign, with a written report on the points to clarify. View the service Example of training programmes by role, from the board to the people who use the tools. AI AI training Programmes by role, from the board of directors to the people who use the tools. View the service

04 · copy the link to this sectionIn everyday use

What does AI need to work for real, on real data?

Five aspects have to be tackled together. Neglect one and the problem moves further down the line, where it costs more.

1 Data The documents that count are agreed, and outdated ones are removed. An assistant that finds two versions of the same procedure may cite the old one.
2 Connections The system works on the ERP, databases, archives and email the company already has. It uses dedicated access, and testing stays separate from the systems in use.
3 Security Permissions are kept to a minimum and requests are logged. We also check that the assistant does not follow instructions hidden in an email or a file (prompt injection).
4 Rules Every system has an owner. There is also a list of the AI systems in use and of the actions that need a person’s confirmation.
5 Over time Answer quality is measured at regular intervals, costs are kept under control, and model updates are tested before use.

05 · copy the link to this sectionHow we work

From the first conversation to everyday use

Each phase ends with a document and a decision by the company.

First conversation

We look at the goal, the systems and the constraints. Is there work to do, and how big is it?

Initial assessment

One or two days with the people who use and run the systems, on site if needed.

Written proposal

Phases, timescales and costs, what is included and excluded, and the criteria for saying it works.

Phased delivery

First in a test environment, then with a pilot group.

Ongoing use

Strategic runs it, with support times and terms written into the contract. Or it passes to your in-house team, with hands-on support.

What does it cost, and how do we start?
There is no price list. The cost depends on how many systems need connecting and on the state of the data. The first conversation carries no commitment. If there is work to be done, you receive a written proposal for the initial assessment, usually one or two days of work with the people involved. After the assessment comes the proposal for the project, with phases, timing and costs.
How long does it take?
The proposal sets the duration and the progress milestones. We start with one area or one process, and widen the scope once the pilot group meets the criteria.
What do we keep?
Code, configuration, the instructions given to the models and documentation, as set out in the proposal.
What if it doesn’t work?
If the pilot group does not meet the criteria, the system is not extended. We decide together how to fix it.

Training programme

06 · copy the link to this sectionCompany Brain

The manual no one can find, last year’s quote, the figure in the ERP

With the Company Brain, the company’s own searchable memory, your people ask a question and get an answer that points to the source document or data. They see only what their department’s permissions allow.

We start with one area, with an owner who decides which documents count. The other areas are added on the same foundation.

The Company BrainAI agents and integrations

Data stays where the company decides. Location and providers are set out in writing in the project.

How it is built

01SourcesThe archives of the first area, current documents only.
02PermissionsSet by department.
03AnswerWith the source document.
04LogQuestions and sources used.

07 · copy the link to this sectionSecurity and rules

The rules that apply to an AI assistant in use

An assistant that reads company documents and data is a new user of your systems. Its rules are written before it starts.

  • AI Act. Organisations that use an AI system in their business must prepare the people who use it (since 2 February 2025). They must also say when someone is talking to an assistant (since 2 August 2026). Stricter rules apply from 2 December 2027 to AI used to select staff or assess creditworthiness.

  • Personal data. The GDPR applies. When the risk to individuals is high, a data protection impact assessment is required, carried out with the data protection officer (DPO).

  • NIS2. Companies within the scope of NIS2 also have to manage the risks posed by assistants connected to their systems.

This is general information, not legal advice. Each case should be checked with the company’s lawyers. For companies in San Marino, the framework has to be assessed case by case.

The full timeline with sourcesNIS2 in plain English

08 · copy the link to this sectionThe team

Who works on the projects

The written proposal states who works on the project and in what role. Client names and the details of their systems remain confidential.

LAB Development and integrations Code is reviewed and tested before every release. DBA Databases ERP data is read without putting load on the systems in use. CYB Security Permissions, logs, and checks on the assistants and on the model providers.
AI A single point of contact They coordinate the team and report to management.

Artificial intelligence in our other areas. We test assistants in cybersecurity and use Oracle data as a source in database work. AI supports cash forecasting in restructuring and foreign-language contracts in international expansion. In Blackroom, by choice, there is no AI.

09 · copy the link to this sectionWhat we don’t do

When we are not the right choice

We tell you in the first conversation when:

  • you need a demo to show, not a system to use
  • you expect firm results on a fixed date, before anyone has seen the data and systems
  • no one in the company will be accountable for the system once it is live
  • the tool you already license does the job with the right settings
  • you want someone to endorse a decision already taken, not to review it
The real test of an AI system is not the demo. It is the third month of use, when the documents, the people and the model change.

10 · copy the link to this sectionFrequently asked questions

Frequently asked questions about AI consulting

What do Strategic’s AI experts do?
They bring artificial intelligence into an organisation’s systems and keep it running. They work out where it is useful, connect it to the ERP and archives, look after its security and rules, and prepare the people who will use it. The same team includes both the advisers and the people who build the connections.
How much does an AI project cost?
It depends on how many systems need connecting, the quality of the data, the security required and who will run the system afterwards. The running costs of the models count too. There is no price list. The first conversation carries no commitment. If there is work to be done, you receive a written proposal for the initial assessment, usually one or two days of work with the people involved. After the assessment comes the proposal for the project, with phases, timing and costs.
How long does it take?
It depends on the systems to connect and on how available the data is. The project proposal sets the duration and the milestones. We usually start with a single area, so the first result comes early and decisions can be based on it.
Do you work with our IT department or instead of it?
With your IT department. Access and go-lives follow your internal procedures, and connections are tested first in a separate environment. At the end of the project, your in-house team can take over the system or hand it to Strategic, under terms written into the contract.
Where is the data processed? Which is better, the cloud or our own servers?
Where the data is stored and processed is decided with the company’s IT team and written into the project. As a rule, we choose services that process data within the European Union. The decision follows explicit criteria. These include how confidential the data is, what the vendor’s contract says about training and retention, and volumes and costs. The two approaches can coexist on different data.
Do you recommend your own products?
When one of Strategic’s tools, such as DIREKTA, Kristina Reply or Autochatter, is among the options, we say so and compare it with the alternatives on the same criteria. In a second opinion on AI projects and vendors, we do not put forward our own products. How we use artificial intelligence

11 · copy the link to this sectionObservatory

From the Observatory: artificial intelligence

Analysis and guides by the Strategic editorial team.

All articles on artificial intelligence

The essential parts of this page print as a short summary of a few pages, without images or invitations to get in touch.

12 · copy the link to this sectionContact

An AI project to set up, unblock or review?

Write a few lines about your company, the systems involved and your goal. We reply by email, usually within two working days.

Or write to us:WhatsAppstaff@strategic.sm

On WhatsApp, messages are answered by Strategic’s AI assistant. Please do not use it for confidential information.

Your requestAIArtificial intelligence

Do any of these questions sound familiar? tick the ones that apply to you

The form opens with the topic and the questions already filled in.