AI Chatbots

Chatbots that do not improvise with your reputation

Every response is pre-validated by humans before it reaches your public. Our deployed chatbots answer customer, student and patient questions without improvising.

  • 96% of questions resolved
  • 50,000+ conversations handled
  • 35,000+ unique users
  • 0 reputational incidents
Book a discovery call

30-60 min. We see if AI is the right tool for your organisation.

AI chatbots for organisations
The technology

Three kinds of chatbot. The choice matters.

Most vendors do not tell you what they are actually building. We choose the architecture together with you, because each one trades flexibility against control differently.

Decision tree

Clear, human pre-validated responses on defined scenarios. Full control over the information and complete predictability. The limitation: it only answers what it was built for.

Right where information accuracy is critical: healthcare, public institutions.

Generative AI

Flexible and adaptive, it handles a wide range of natural-language questions. The trade-off: responses can be unpredictable, and control over content is reduced.

Right where flexibility matters more than strict control.

Hybrid

The validated structure of a decision tree plus the natural interaction of generative AI. A balance of safety and flexibility, with the knowledge base under control.

Right for most organisations that want both: safety and natural conversation.

Comparison of chatbot architectures
Capability Decision tree Generative AI Hybrid
Available 24/7 ✓ ✓ ✓
Human pre-validated responses ✓ ✗ ✓
Predictable and consistent answers ✓ ✗ ✓
Control over content ✓ ✗ ✓
Flexible with unforeseen questions ✗ ✓ ✓
Hallucination-resistant ✓ ✗ ✓
Reputational risk very low high controlled
The safety methodology

Why there have been no incidents

  • Pre-validated responses. Content is verified by humans before it enters the chatbot.
  • Controlled reputational risk. The chatbot does not improvise on sensitive topics; on uncovered scenarios it routes the user to a human.
  • Rigorous testing before launch. Every chatbot goes through real scenarios with real users before going public.
  • Continuous monitoring and updates. We track unanswered questions and keep the knowledge base current.

The methodology comes from direct experience in regulated environments: the founder contributed to operational chatbots within the European Central Bank's Digital Transformation and AI Programme.

The training programmes were delivered during the founder’s time working at the European Central Bank. This does not necessarily constitute a public endorsement of AI Solutions or its services by the European Central Bank.

Deployments

Chatbots in production

Public university, Romania

Faculty of Economics and Business Administration, Babeș-Bolyai University of Cluj-Napoca

Instant answers to student questions, 24/7

Admissions, registrar matters, and frequent questions about study programmes.

The Ask FSEGA chatbot in use, answering a student's question about study programmes (interface in Romanian)

Public university, Romania

Faculty of Economic Sciences, University of Oradea

Admissions and registrar information, without the queues

The second university chatbot built on the same validated methodology.

The University of Oradea economics faculty chatbot in use (interface in Romanian)

Private oncology clinic, Romania

Medisprof Cancer Center, Cluj-Napoca

Patients get immediate answers to frequent questions

The first project in a partnership that grew into a full AI adoption programme.

The Medisprof chatbot in use, answering patient questions (interface in Romanian)

The FSEGA chatbot experience, in the client's words

The video loads from YouTube when you press play.

Let's see what a chatbot for your public would look like

A 30 to 60 minute conversation about your needs. We recommend the right architecture, with the reasoning on the table.

Book a discovery call

30-60 min. We see if AI is the right tool for your organisation.