AI in your company
ChatGPT integration for your company, from team plans to AI inside your tools
We help you choose the right plan and model, set privacy correctly, train your people and connect AI to your CRM, helpdesk or company documents.
Three levels of AI in a company
Bringing ChatGPT into a company happens on three levels: team plans set up correctly, rules and training for your people, and API integrations into the tools you already use. The first two usually come before the third. We do not resell subscriptions; we help you choose, configure and build. An integration plan costs €1,200–2,500, AI added to one workflow €400–1,200, and an AI feature inside your company's software starts at €4,000.
The options, from simple to complex
| Option | What it means | Cost | Good for |
|---|---|---|---|
| Team plans | ChatGPT Business or Enterprise, Claude Team or Enterprise, Gemini in Google Workspace, Microsoft 365 Copilot, with central admin | Per user per month; check the vendor's current price | Writing, summaries, analysis, research |
| Assistants with company documents | Projects or assistants set up inside the team plan, with procedures, templates and internal guides | Included in the plan, plus setup time | Internal answers, quotes, standard documents |
| API integration | The model is called from your CRM, helpdesk, ERP or app, with your data and rules | Pay as you go, per token | Repetitive processes, high volumes, external customers |
Privacy differs from plan to plan
On personal accounts, OpenAI says it may use conversations for training unless you switch off the model-improvement setting. For ChatGPT Business, Enterprise, Edu and the API, it says data is not used for training by default. That difference is the main reason a company pays for a business plan instead of leaving each employee on a personal account. Other services are compared in our guide to AI chatbot privacy settings. On any plan, passwords, API keys and card details do not belong in a chat.
Staff training and the AI Act
Article 4 of the AI Act asks companies that use AI to take measures supporting the AI literacy of the people who work with those systems. The duty has applied since 2 February 2025, and national authorities supervise it from 2 August 2026. The European Commission says no certificate is needed and that you can keep an internal record of training. We train people on their real tasks: which data they may put into a chat, how to check an answer, where models often go wrong and when a text must be labelled as AI-generated. More in our AI Act guide for small businesses and our guide to using AI well.
Which model to choose
ChatGPT is one option among several. Claude, Gemini and Mistral, a European company, offer comparable business plans, and open-weight models can run on your own servers when data must not leave the company. We choose by task: a small, cheap model for classification or extraction, a more capable one for long texts and reasoning. Before deciding we test on 30–50 real examples from your business, because a public leaderboard does not show how a model handles your documents.
What you pay, and to whom
Subscriptions are paid directly to the vendor, per user, and the API by usage. We invoice the work: the integration plan with an evaluation set, the automations and the AI features inside your tools. Training is priced in the quote, by number of people and roles. To see where Romanian companies stand, read our analysis of AI adoption in Romanian SMEs.
What we set up
AI your people can use safely
From the first subscription to AI built into your own tools.
Team plans set up properly
We pick the plan and configure admin, access and data settings, so staff stop using personal accounts for work.
Privacy settings
Training switched off where relevant, retention defined, connectors enabled only where needed, and a clear list of what never goes into a chat.
Training on real tasks
Sessions built on each role's actual work, with an internal record for the AI literacy duty in the AI Act.
Internal AI usage policy
A short document: approved tools, which data may go in, who checks the output and how generated content is labelled.
The right model for each task
We compare models on your own examples, including European options and local models for sensitive data.
API integration
AI inside your CRM, helpdesk, ERP or intranet, with minimal data sent, logs and per-request cost tracking.
How it works
How we bring AI into a company
- 1
Usage audit
We find out who already uses AI, on which accounts, for which tasks and with which data. Personal accounts used for work surface here, if there are any.
- 2
Policy and tool choice
We settle the plan, privacy settings, allowed data categories and who approves new tools.
- 3
Training
We train each role on its own tasks and leave an internal training record, as the European Commission suggests.
- 4
Integrations and measurement
Where a process repeats, we connect the model to your tools through the API and measure time saved against the baseline.
Budget
AI integration prices
Our fees cover the work only, are indicative and negotiable, and are agreed in writing. Subscriptions and API usage are paid directly to the AI vendor.
AI integration plan
€1,200–2,500one-off
An audit, a model choice tested on your data and a written plan.
- Audit of current AI use
- Evaluation set built from your tasks
- Model and cost comparison
- Internal AI usage policy
AI in one workflow
€400–1,200per workflow
The most common first project: one repetitive task handed to a model, with checks.
- One repetitive process
- Model chosen on cost and quality
- Human approval where it matters
- Documentation and handover
AI feature in your software
from €4,000one-off
AI built into the systems your staff or customers already use.
- API integration into CRM, helpdesk or ERP
- Answers grounded in company documents
- Logs, cost limits and monitoring
- Evaluation before launch
AI tools we have built
My Claude Code is a local LLM gateway we built: one address for 57 model providers, with tiered routing, fallback chains, the cost of every request and keys kept in a single local file. We apply the same thinking when choosing models for a company. AI Translator Pro, in development, translates Word, PowerPoint and PDF documents with local or cloud models chosen per job, and AI Transcription Studio, also in development, transcribes on your own machine so the recording never leaves it. For new projects, see our AI systems service.
FAQ
Questions, answered
Is it safe to use ChatGPT with company data?
It depends on the plan and the settings. OpenAI says ChatGPT Business, Enterprise, Edu and the API are not used for training by default, while personal accounts may be unless the user opts out. Even on a business plan, send only the data a task needs and never paste passwords, API keys or card details. A short internal policy makes this clear to everyone.
What is the difference between ChatGPT Plus and ChatGPT Business?
Plus is a personal subscription tied to one person's account. Business is a company workspace with central admin and billing, where OpenAI says data is not used for training by default. When someone leaves, their work stays with the company. Features and prices change often, so check OpenAI's current plan page before you buy.
Do we have to train employees under the AI Act?
Article 4 requires companies that use AI to take measures supporting the AI literacy of the people who use it on their behalf. The duty has applied since 2 February 2025 and is supervised from 2 August 2026. The Commission says no certificate is required and an internal record of training or guidance is enough. Penalties are set by national law.
How much does it cost to integrate ChatGPT into a company?
There are two kinds of cost. Subscriptions are paid to the vendor per user per month, and API usage is paid per token. Our work is separate: an integration plan with an evaluation set costs €1,200–2,500, AI in one workflow €400–1,200, and an AI feature inside your software starts at €4,000. Training is priced by number of people and roles.
Should we use ChatGPT or another model?
It depends on the task. ChatGPT, Claude, Gemini and Mistral each do some jobs better or cheaper, and a small model is often enough for classification or extraction. We test candidates on 30–50 real examples from your business and compare quality and cost. Building integrations so the model can be swapped keeps you free to change when prices or quality shift.
Can we use AI without data leaving the company?
Yes, with open-weight models running on your own server or computers. They work well for transcription, classification, extraction and drafting from internal documents. The trade-offs are hardware cost, maintenance and lower quality than the largest cloud models on the hardest tasks. A common mix is local models for sensitive data and a cloud model under business terms for the rest.
What does API integration actually mean?
Your software sends a request to the model with the data and instructions for one task, and the answer comes back into the same software. A helpdesk can draft a reply inside the ticket, a CRM can summarise a call, an ERP can read a supplier's PDF. Nobody copies text into a chat window, every request is logged, and you pay only for what is used.
How do we know whether AI is actually helping?
Measure before you change anything: how long a task takes, how many cases come in, how many errors slip through. After a month with AI, measure the same things and subtract the subscription, usage and review time. Our AI agents ROI guide shows the full calculation with a worked example.
Make a plan for AI in your company
In a free call we look at how your people use AI today and what is worth changing first.