Should every business have its own AI?
· 6 min read · TurkİSG AI
Everyone has a chatbot at hand. You ask, it answers. But that bot does not know your clinic's opening hours, your OSGB's list of client companies or your real estate office's portfolio. It talks with general knowledge, not with yours.
A business's own AI closes exactly that gap.
General bot vs. business intelligence
A general bot: Trained on the general knowledge of the internet. It gives a reasonable answer to every question and the right answer for you to none of them. It has no answer to "do we have an appointment tomorrow?".
The business's own AI: Works with your data. It knows your appointment book, your patient records, your regulation archive, your portfolio. It says "there's a free slot tomorrow at 14:00, and Ms. Ayşe's check-up is on the same day". It cites its sources. If it gets something wrong, you correct it and it learns.
Who needs it?
If two of these signs are present, the time has come:
- The same questions are answered many times a day (appointments, prices, opening hours, regulations).
- Information is scattered: Excel, WhatsApp, e-mail, folders. Finding something takes longer than asking.
- There is a single expert; when they are away, work waits.
- Requests that arrive after hours wait until the morning.
- Reporting is done by hand and the same effort is spent every month.
How is it built?
A business AI does not mean training a model from scratch. It has three layers:
- Data. The business's own knowledge is put into an organised structure: documents, records, process rules. Without this layer, the AI talks without substance.
- Intelligence. A ready-made language model is connected to the business's data (RAG — retrieval-augmented generation). The model knows language in general, produces the answer from your data and shows its source.
- System. The AI does not stay a chat box; it books appointments, writes corrective and preventive action reports, answers customers on WhatsApp. It becomes part of the workflow.
An example: TurkİSG AI
The assistant we built for occupational health and safety professionals works with more than 91 regulation documents and 89 guides from the Turkish Ministry of Labour and Social Security (ÇSGB). It answers regulation questions by pointing to the article, produces a hazard analysis and a corrective and preventive action report from a site photo, and lets you practise with more than 2,700 exam questions. A general bot can do none of this, because the data is not in it.
The same approach is adapted to a clinic, a real estate office, an OSGB. The data changes; the method stays the same.
The critical condition: ownership
A business's AI works with the business's data. If that data goes to someone else's server, the AI becomes someone else's too. In the right setup, the data stays in your own infrastructure, the model reaches it there, and nothing leaves.
If the AI is learning the business's memory, that memory should stay in the business.