As artificial intelligence evolves, clinicians and healthcare leaders face a critical question: Will one all-powerful AI dominate, or will we rely on a constellation of smaller, specialised systems working together in harmony?

From my experience as a clinician and educator, the answer is clear. We will not entrust decisions to a single "Dr. AI." Instead, we will collaborate with a network of focused, expert AI assistant, each solving specific clinical challenges with precision, safety, and transparency.

While general-purpose models like GPT-4 demonstrate impressive versatility, healthcare demands an unmatched level of rigour. Precision, explainability, and trust are non-negotiable. That is why specialised AIs, trained and validated for individual tasks, are critical. For instance, radiology tools identify subtle findings on scans, symptom triage bots help shape patient flow, and tools like Katana AI enhance prescribing safety by flagging drug interactions, highlighting adherence issues, and suggesting dose adjustments in real-time.

The future of healthcare AI will not be monolithic; it will be modular. Like the brain with its specialised lobes, clinical AI will integrate discreetly into EHRs and clinician workflows. Each module excels at a specific function, then shares insights via smart, secure APIs.

In practice, clinicians will not "use" AI; it will be ambient, helping in the background. It quietly drafts documentation during patient conversations, alerting you to prescribing errors before they occur, or reminding you of critical guidelines that align with standard care.

With such capabilities comes responsibility. The most trusted medical AI platforms will be traceable, clinically validated, and compliant with regulations. Tools like Katana AI earn their place by providing clear audit trails, evidence-based decision support, and updates in line with emerging research, far more accountable than any single, opaque mega-model.

The best AI systems will also be co-trained with clinicians. By learning from real-world practice patterns and sometimes adapting to local hospital protocols, they become contextually intelligent, not just broadly knowledgeable.

Looking ahead, expect purpose-built AIs across radiology, prescribing support, triage, referrals, documentation, and beyond, each connected but independently optimised. Doctors remain at the helm. AI does not aim to replace us—it is here to empower us, effortlessly catching errors, streamlining tasks, and enhancing patient care.

So, no, there will not be one super AI. Instead, we will have many small, brilliant tools, like Katana AI for prescribing safety, integrated seamlessly into everyday clinical work. The doctor stays in charge, and AI becomes our most trusted assistant.

Mr Derek Hennessey MD, PhD, FRCS(Urol) FEBU Consultant Endourologist and minimally invasive surgeon