“AI doesn’t just follow orders – it takes the initiative”

To date, the greatest impact of AI adoption in optometry has centred around analysing retinal images.

However, we are now entering an era where AI doesn’t just follow orders –it takes the initiative. This new wave, known as agentic AI, refers to software entities that can set their own goals, make decisions, and perform tasks autonomously on behalf of humans. These AI agents act as independent digital assistants.

The reduced adminfor optometrists means that practices can serve more patients. This is exciting for practice owners and multiples that are keen to improve operational efficiency under increasing financial pressures

Across the pond

Today, in opticians practices across the US, you can walk into a testing room and have an AI agent that instantly analyses optical coherence tomography (OCT) scans, flags potential pathologies, and prepares patient notes, just seconds after the patient has been pre-screened.

Companies like Altris AI have developed AI to read OCT B-scans, colour-coding areas of pathology with nearly 90% accuracy: red for urgent cases like wet age-related degeneration (AMD), yellow for moderate concerns, and green for normal results. This allows an AI agent to decide who needs an urgent referral and to draft the referral letter in seconds.

Other AI tools, like Barti and NextGen Ambient Assist. allow for more eye contact with patients by transcribing conversations in real time, summarising history and symptoms within 60 seconds, and claiming to save optometrists up to two hours of documentation time daily.

The reduced admin for optometrists means that practices can serve more patients. This is exciting for practice owners and multiples that are keen to improve operational efficiency under increasing financial pressures and will therefore push for rapid AI adoption.

However, practitioners I speak to are concerned about increased litigation risks associated with AI tools, and prefer a more cautious approach to integrating them into their daily workflows. This is one of the reasons why AI solutions have yet to be integrated by providers of electronic health records in the UK.

I believe every optometrist should consider taking a course on data and AI fundamentals: the more you understand how AI works, the better you will be at recognising its limitations and integrating it safely into your practice

The future in fast forward

But change is coming fast. Later this year, it will be possible to interact with AI that doesn’t just flag conditions, it will explain its “chain-of-thought” reasoning in detail.

Instead of merely presenting a diagnosis, an AI agent will be able to say, “I see drusen at the macula that are 63% similar to a known risk pattern for dry AMD. I recommend daily lutein supplements, weekly monitoring of Amsler grid and a follow-up in six months.” This kind of explanation will help us understand the AI’s logic and make better decisions with it.

Future AI systems will combine information from different tests, such as OCT scans, visual field tests, genetic screenings, and even data from smart contact lenses that monitor ocular pressure. Early prototypes of these multimodal AI systems are already showing promising results, especially in predicting glaucoma more accurately.

In my opinion, it is inevitable that every staff member in an optical practice will work with AI systems and will need some level of data and AI literacy to understand them. I believe every optometrist should consider taking a course on data and AI fundamentals: the more you understand how AI works, the better you will be at recognising its limitations and integrating it safely into your practice.

Of course, there are challenges. What happens if an AI’s decision is biased? Who is accountable if the diagnosis is wrong?

About the author

Jess Jeetley MBE is an optometrist, certified AI management practitioner and AI strategy advisor at Techstars, leading teams through ISO 42001 implementation and building AI literacy programmes.