BDJ In Practice spoke to Hannah Burrow and Jay Shah, Co-founders of Kiroku, to discuss all things artificial intelligence.

What are the biggest advantages of AI you have seen implemented within dentistry?

JS AI's clearest win so far is taking routine, time-heavy work off clinicians' shoulders. Real-time, AI-powered note-taking systems now transcribe, structure and code clinical records as we speak, so the dentist can concentrate on the patient instead of the keyboard. Because the notes are generated consistently and mapped to standard terminologies, compliance checks, referrals and audits run faster too.

For individual practitioners, that means less cognitive load, the chance to leave on time, and if they wish, finishing any remaining admin from home rather than staying late in the surgery.

AI is also becoming a reliable second pair of eyes in diagnosis. Image analysis tools that highlight early dental caries, infection around tooth roots or the first signs of gum disease are already embedded in many practice management systems, providing immediate feedback while the radiograph is still on screen.

HB Equally tangible and very patient-facing is the latest generation of AI-assisted intraoral scanners. As the wand moves around the mouth, the software builds and refines a 3D model in real time, automatically cleaning up artefacts and suggesting preparation margins. Patients can literally watch their own teeth appear on the monitor, which is a powerful aid to communication and treatment acceptance.

Taken together, these technologies standardise data capture, speed up clinical and administrative workflows, and make expertise more scalable across a group of practices. The result is a positive cascade: practices run more efficiently, dentists enjoy a lighter workload and better work-life balance, and patients receive more personalised, dependable care with fewer surprises and shorter chairside times.

Conversely, what are the biggest dangers of using AI in dentistry?

JS The chief risk is over-reliance on technology without understanding its limits. AI systems are trained on data that is often fragmented, inconsistent or biased; if those flaws creep into the model, they can lead to missed or incorrect diagnoses.

Another danger is the ‘black box' problem. If the algorithm's reasoning is opaque, the dentist cannot justify a clinical decision to a patient, undermining both professional judgment and trust.

HB Expectation management matters here. People increasingly assume that ‘AI' equals perfection, yet these tools are iterative: models are constantly updated and refined, and in many cases, improving. Clinicians and patients alike need to treat today's output as a best-available estimate, not gospel truth, and be prepared for results (and user interfaces) to evolve with time. Data privacy adds another layer. Dental records are deeply personal, and every AI vendor must meet stringent standards for security, auditability and regulatory compliance, especially as new entrants with little healthcare heritage appear.

In short, AI should augment, not replace, human expertise. The danger arises when it becomes a shortcut: when clinicians skip the contextual, empathetic elements of care or delegate final judgment to software that is still a moving target.

Which areas have developed the most?

JS One of the most significant areas of progress has been in efficiency and documentation. AI-driven clinical note generation has accelerated rapidly. Systems can now listen in real time, transcribe, structure, and in some cases, even code the clinical encounter. This allows dentists to finish appointments with a complete, standards-compliant record instead of a backlog of unfinished notes. The gains are tangible: minutes saved per patient, fewer late nights, and a substantial reduction in repetitive admin.

Diagnostics have also come a long way, particularly in tools that clinicians can use directly with patients. Deep-learning algorithms that analyse radiographs have become a reliable safety net, highlighting subtle caries, early periapical lesions or crestal bone loss that tired eyes might miss. Many platforms now overlay colour-coded markers directly onto the x-ray, allowing the dentist to point to the screen and say, ‘Here's what I'm concerned about' - making the diagnosis more concrete and easier for the patient to understand.

HB Another major shift has occurred in visualisation within the mouth. AI-assisted intraoral scanners have evolved from novelty to near-standard tools in restorative dentistry. As the scanning wand moves, the software builds and refines a 3D model in real time, removing artefacts and even suggesting margin lines within seconds. Being able to show patients a live, rotatable model of their teeth makes issues like cracks, wear facets or crowding instantly obvious, often communicating more clearly than words or flat 2D images ever could.

Communication with patients is also being transformed. Large language models are now being used to generate personalised post-operative instructions, translate explanations automatically, and deliver behaviour-change nudges in plain language. Instead of receiving a generic leaflet, patients can now get a tailored message that refers to their specific restoration, their next hygiene visit, and even includes an image from their scan - boosting understanding, trust and long-term adherence.

Taken together, these advances represent a real shift from experimental prototypes to practical, everyday tools. They streamline practice operations, strengthen diagnostic confidence and, perhaps most importantly, open a window that lets patients see what the dentist sees.

Do you think there is - or could be - a generational divide between those who are keen to use AI, those who are cautious and those who flat refuse to?

HB There's certainly a spectrum of enthusiasm, caution and resistance toward AI, but it splits more by mindset and available time than by age. Younger clinicians, used to apps and instant feedback, are often more comfortable investing time upfront to customise templates or settings. With support from onboarding teams, they quickly see the payoff in shorter appointments and fewer late-night admin sessions.

Across all age groups, you also find time-pressed pragmatists: dentists who'll adopt any tool that clearly saves minutes, as long as the setup is simple and the support is there. The difference isn't age, it's bandwidth and trust that the benefit will be immediate.

JS Sceptics often raise concerns about de-skilling, compliance or data privacy. But when AI offers transparency, like showing why it flagged a lesion, and gives clinicians clear editorial control, attitudes shift. Many who were initially resistant soften once AI is built quietly into the software they already use. When it becomes a feature, not a buzzword, objections fade.

Institutional silence or, on the other hand, recommendations without consulting the experts in the field can amplify anxiety. That's why regulators, educators and providers like the GDC, HEE and NHS need to engage with providers as well as users regularly and early. If the profession co-designs standards for safety and accountability, we can shape these tools from the ground up. While age affects tech familiarity, the real drivers of adoption are time, trust, and transparent support. With the right approach, enthusiasm for AI can span every generation.

In your view, what areas are ripe for development and integration of AI, and which areas should we be most wary of?

HB One of the ripest areas for AI integration is still clinical documentation and admin support. Dentistry is full of repetitive, time-consuming tasks; notetaking, formfilling, referral letters, triaging medical histories, that take up clinical time. AI can meaningfully reduce this burden, freeing clinicians to focus on patient care.

There's likewise huge potential in personalised patient communication. Language models can tailor treatment plans, reminders and explanations to a patient's reading level or preferred language, boosting understanding and trust, especially for those with different communication needs.

JS On the clinical side, decision support tools such as radiograph analysis, charting aids and risk prediction algorithms are becoming more robust. The real test is integrating them so they enhance, rather than override, professional judgment.

The hazards cluster around opacity, bias and data stewardship is one area that all AI models need to get better at. Blackbox systems that cannot justify their suggestions erode confidence, invite legal risk and weaken patient trust. At the same time, largescale data collection demands clear safeguards on privacy, consent and equitable model performance.

Ultimately, AI in dentistry must be assistive, transparent and ethically grounded, not merely impressive on paper.

How are you maximising the capabilities of AI?

JS We use AI across the business, not just in the product itself. It's part of how we build, operate and support our customers. Our product and engineering teams use AI to design and develop faster, but also more creatively. Whether it's generating UI concepts, automating testing, or rapidly iterating on prototypes, AI helps us move from idea to implementation at speed.

This acceleration means we can build and ship high-quality features much faster than would have been possible even a year ago. It helps us maximise the impact of our team, delivering more value faster, without compromising on quality or creativity.

HB Operationally, we use AI to automate repetitive admin tasks, streamline internal processes and reduce cognitive load for the team just as we do for clinicians. And on the customer side, AI helps us understand users better: surfacing common pain points, tailoring support, and delivering more timely, relevant help.

Ultimately, AI helps us stay agile, lean and focused on what matters - giving clinicians better tools, faster, and making their working lives easier.

Is AI in dentistry a friend or a foe?

JS AI is absolutely a friend when used responsibly. Like any powerful tool, its value depends on how it's designed, implemented and used. When developers build with clinical context, transparency and safety in mind, and when clinicians treat AI as an aid rather than a crutch, it becomes a force multiplier: saving time, improving accuracy, and enhancing patient care.

HB But it can become a foe if misused. For developers, that means building black-box tools without clinician oversight, prioritising speed over rigour, or treating sensitive health data carelessly. For users, the danger lies in over-reliance: trusting outputs blindly, skipping critical thinking, or assuming AI can replace rather than support professional judgement. The best results come when both sides embrace AI with clarity and accountability. If we build it thoughtfully and use it wisely, it's not just a friend - it's an essential partner in modern dentistry. ◆

figure 1
figure 2