A Cape Town med-tech company has taken a major step in the fight against tuberculosis, with AI Diagnostics raising R85-million in a pre-series-A funding round to scale an AI-powered stethoscope built to spot TB earlier and closer to the patient. The raise is another sign that South African health-tech is beginning to attract serious backing for technologies aimed at some of the continent’s toughest public-health challenges.
The company’s pitch is simple but ambitious: use artificial intelligence to help front-line health workers identify TB signals in lung sounds in real time, even in settings where X-rays, specialist doctors and lab infrastructure are limited. For communities where people often present first at clinics, pharmacies or through community health workers, that could make a meaningful difference in how quickly patients are referred and treated.
The funding round was led by the Steele Foundation for Hope, and it will go toward rolling out Ostium, the company’s digital stethoscope, alongside AI.TB, its proprietary model trained to recognise patterns associated with tuberculosis. In practical terms, the technology is designed to give nurses and other primary healthcare workers a diagnostic assist at the point of care, instead of waiting for results from more resource-intensive systems.
AI Diagnostics CEO Braden van Breda says the company deliberately chose TB because it remains one of the world’s most underfunded and stubborn infectious diseases. That choice is especially relevant in South Africa, where TB continues to place pressure on the public health system, particularly in high-burden communities such as parts of the Western Cape, Gauteng and KwaZulu-Natal.
Van Breda’s argument is that machine learning can hear what human clinicians may miss. The company says its model has been trained to detect subtle TB-related audio cues in the lungs, something that requires not only sophisticated software but also a carefully built dataset of TB-positive lung sounds. That dataset, the company says, had to be created from scratch, making the technical work as much about medicine and field collection as about coding.
As we reported earlier in our coverage of the local health-tech space, the real test for products like this is not whether they impress in a demo room, but whether they can survive the chaos of everyday clinics. AI Diagnostics appears to know that. The company has spent time dealing with noisy environments, paediatric cases, and the complexity created by conditions such as HIV co-infection, which can complicate TB screening in South Africa and across the region.
AI Diagnostics and the push for TB screening at scale
The AI Diagnostics model is being positioned as a point-of-care screening tool that could be used by nurses, pharmacists and community health workers — the people most likely to see patients first in low-resource settings. That matters because the earlier TB is picked up, the sooner a patient can be tested properly and started on treatment, helping reduce transmission in homes, workplaces and schools.
The company is also grappling with a challenge that often gets overlooked in AI health stories: how to stop a model from drifting or degrading when it leaves the environment in which it was trained. A device that performs well in Khayelitsha might not work identically in rural Zambia or Vietnam, where clinic conditions, patient populations and background noise can differ substantially.
That is why the deployment story is just as important as the algorithm. If AI Diagnostics wants to prove this technology can work beyond a handful of pilot sites, it will have to show that Ostium can remain reliable across different countries and care settings, while still giving clinically useful results to non-specialist users. That will be key to winning trust from health systems that are often wary of promising tech that cannot hold up under real-world pressure.
Another major hurdle is regulation. Van Breda has spoken about the company’s ambition to secure World Health Organisation certification, a milestone that would strengthen its credibility with governments, donors and health agencies. For health-tech startups, especially those dealing with diagnostic tools, certification is not just a box-ticking exercise. It is often the difference between a promising innovation and a tool that health departments are willing to buy, deploy and support.
The company is also aware that it is not operating in a vacuum. AI-assisted chest X-ray systems such as CAD4TB and Qure.ai’s qXR already play a role in TB screening workflows in some markets. The question is not necessarily whether one tool will replace the other, but how these technologies can be used together to widen the net and identify more patients earlier.
For South Africa, where public clinics are often overcrowded and diagnostic pathways can be slow, the possibility of an affordable, portable AI-assisted screening tool is appealing. But the scale challenge is huge. Building something that works is one thing; getting it into every primary healthcare clinic on the continent by 2030 is another matter entirely.
That target would require not only manufacturing and logistics, but procurement partnerships, clinician training, regulatory approvals and sustained funding. It would also require national health systems to believe that the device can help reduce missed cases without creating unnecessary referrals or extra workload for already stretched staff.
The broader significance of the funding round is that it shows investor appetite for African health innovation remains alive, particularly where solutions are tied to conditions that disproportionately affect local populations. Unlike some imported digital health products, AI Diagnostics is tackling a problem that is deeply familiar on the continent and building from a South African base.
For now, the company’s progress offers a glimpse of where medical technology may be headed: smaller devices, smarter software and more diagnostics pushed closer to the patient. Whether Ostium becomes a routine part of TB screening across Africa will depend on evidence, regulation and rollout at scale. But with R85-million now in hand, AI Diagnostics has given itself a far better chance of turning a bold idea into a public-health tool that could matter on the ground.