Oct 5, 2025
6
min read
Medically Reviewed
Share
The Capacity Test: A Simple Framework
The capacity test is straightforward but revealing. For any technology under consideration, the practice asks a single question: does this tool create more time and energy for patient care than it consumes in setup, training and ongoing management? A tool that saves the clinician ten minutes per consultation but requires thirty minutes of training per week and a daily login to a separate system may have a net negative effect on capacity. A tool that saves two minutes per call but requires staff to manually enter data from the tool back into the practice management system is not creating capacity at all.
The test is not about whether the tool has impressive features or whether it represents cutting-edge technology. It is about the net effect on the practice’s ability to deliver care. A simple AI tool that eliminates one repetitive task with zero training and zero workflow change creates more real capacity than a sophisticated AI platform that requires process redesign, team retraining and ongoing manual oversight, even if the sophisticated platform has more technically impressive capabilities.
Practices that apply the capacity test rigorously before purchasing AI tools make better investment decisions. They avoid the trap of buying technology that looks impressive in a demo but creates more work than it saves. They focus their budget on the tools that will have the greatest net positive impact on their team’s capacity. And they create a culture in which technology is expected to serve the practice, not the other way around.
Where AI Goes Wrong: Adding Complexity
AI can add complexity to a practice in several ways, and recognising these patterns is the first step to avoiding them. The most common is the separate-system pattern, in which the AI tool operates on its own platform and requires staff to log into a different interface, learn different navigation and manually transfer data between the AI system and the practice management system. This pattern is especially common among AI tools that were built as standalone products and later marketed to healthcare without genuine integration.
The second complexity pattern is the workflow-addition pattern, where the AI introduces new process steps that did not previously exist. A documentation tool that requires the clinician to tag, categorise or format their dictation in a specific way before the AI can process it is adding steps to the workflow, not removing them. A billing tool that requires the practice manager to run a separate reconciliation process is adding work, even if the billing suggestions themselves are accurate.
The third pattern is the exception-handling burden. AI tools that handle routine cases well but generate errors or require manual intervention for non-routine cases can actually increase workload if the exceptions are frequent. The receptionist who must check and correct every third booking made by an AI receptionist is not saving time; they are doing their original work plus AI oversight. The key metric is not how well the AI performs on average, but how much staff time is consumed by handling the cases the AI cannot resolve autonomously.
Expert Tips
"Every technology a practice adopts should be held to a simple test: does it make the practice's life easier or harder? The unfortunate reality is that many well-intentioned technology investments fail this test because they add new logins, new processes and new data entry requirements on top of the existing workload. AI that creates capacity does not ask staff to learn anything new; it simply makes the things they already do happen faster and more accurately. If you have to train your staff to use an AI tool for more than an hour, the tool is probably adding complexity rather than removing it." — Arash Zohuri, CEO, MediQo
The Design Principles of Capacity-Creating AI
AI that creates capacity rather than complexity follows a consistent set of design principles. The first principle is integration: the AI operates within the systems the practice already uses, not alongside them. It reads from and writes to the practice management system directly, so no data entry is duplicated and no information is lost between systems. The staff member does not need to switch applications or remember to update two systems after a single transaction.
The second principle is autonomy: the AI handles the complete workflow for the cases it can manage, escalating only the exceptions that genuinely require human judgement. The AI receptionist that resolves a routine booking call from start to finish without any staff involvement has created capacity. The AI that handles the first three steps of a booking but requires a human to complete the fourth has only partially addressed the problem, and the partial solution may not be worth the implementation cost.
The third principle is invisibility: the AI does not announce itself with new interfaces, new buttons or new processes. It works in the background, and the only evidence of its presence is that tasks are completed faster and more accurately. Staff do not need to remember to use it; it is simply part of how the practice operates. This invisibility is not an incidental feature; it is the design outcome that determines whether the AI creates net capacity or net complexity.
Key Takeaways
AI should be judged by whether it increases the practice's capacity to deliver care, not by how advanced the technology is.
Many technology implementations fail because they add complexity to workflows rather than removing it.
The most effective AI tools are those that integrate into existing workflows and eliminate steps, not those that introduce new processes.
Practices should evaluate AI investments by asking one question: does this create more capacity than it consumes?
The healthcare sector has a complicated relationship with technology. For every tool that genuinely simplifies a clinical or administrative process, there are several that add steps, require additional logins and generate more data to manage without providing a clear compensating benefit. The cumulative effect of these well-intentioned but poorly integrated technologies is that many practice staff feel they are spending more time managing their tools than delivering care — a problem that has been dubbed 'death by a thousand clicks' and that is a significant contributor to clinician burnout and practice inefficiency.
As AI enters the healthcare technology landscape, there is a real risk that it will follow the same pattern — adding new interfaces, new workflows and new complexity on top of an already overloaded system. The vendors who market AI as a powerful new capability that requires dedicated training and new processes are, whether they intend to or not, repeating the mistakes of the previous generation of healthcare technology. The alternative is to hold AI to a different standard: does it create more capacity for care than it consumes in complexity?
This article argues that capacity creation, not technological sophistication, should be the primary criterion for evaluating AI in healthcare. It examines the principles that distinguish capacity-creating AI from complexity-adding AI, and it provides a framework that practice owners can use to ensure their AI investments deliver the one thing that matters most: more time and energy for patient care.
Share