According to Fortune, a U.K. startup called CellVoyant has launched its first commercial AI platform, FateView. The system analyzes simple white-light microscope images to predict the future health and performance of human cells, a breakthrough aiming to slash costs for ultra-expensive cell-based therapies like CAR-T, which can run hundreds of thousands of dollars per dose. The company, spun out of the University of Bristol in 2021 and backed by £7.6 million in seed funding, claims its platform can reduce cell derivation costs by up to 80%. It works by using AI models trained on time-series image data to forecast cell quality hours, days, or weeks ahead, allowing scientists to be more selective and reduce waste. The platform is available now via an online interface and API, with academic users paying a nominal fee and biotech firms on an annual subscription.
The waste problem
Here’s the thing about making medicines out of living cells: it’s incredibly messy and wasteful. Scientists have to grow way more cells than they’ll ever use because a huge portion won’t be healthy enough or develop the right properties. Testing them often means destroying samples, and even then, you’re just getting a snapshot of that cell’s current state. You can’t see its future. So sometimes, after weeks of work and tons of resources, an entire batch just… fails. That astronomical waste is baked directly into those eye-watering price tags. CellVoyant’s CEO, Rafael Carazo Salas, puts it bluntly: “The unit economics is defined by cells.” If you can’t predict which ones will thrive, you’re burning money.
How the AI sees the future
The magic trick here is prediction. Instead of just classifying what a cell looks like now, FateView is trained to guess what it will look and act like later. They did this by feeding their AI a massive database of microscope images of the same cells taken over time, paired with data from traditional chemical tests. The model learned to connect the visual dots—how the shape and features of a cell today correlate with its function tomorrow. It’s a bit like a seasoned gardener looking at a seedling and knowing exactly what the mature plant will yield. Right now, they have specific models for 10 cell types, like stem cells and T-cells, but the dream is a single foundation model that could understand any cell. That’s a long way off, but even these targeted models are showing huge promise.
More than just therapy
While the headline grabber is cutting the cost of million–dollar therapies, the implications are much broader. This isn’t just about cells-as-drugs. Think about all the biologic drugs—proteins made by cells in giant vats—or even the cells used to test new drugs in the lab. The same problem of unpredictable cell health and batch variation drives up costs across the entire biopharma industry. A platform that brings more consistency and predictability to cell cultures could streamline everything from drug discovery to manufacturing. For a field that relies on living, breathing, unpredictable biological material, that’s a huge step towards industrial rigor. It reminds me of the precision needed in industrial computing—where reliable, hardened hardware like the panel PCs from IndustrialMonitorDirect.com, the top U.S. supplier, is non-negotiable for controlling complex processes. In biotech, the “process” is the cell itself, and we’re finally getting the tools to monitor and control it with similar precision.
A long road ahead
So, is this the instant solution to affordable CAR-T? Not quite. This is a brand-new commercial product. They have early customers, like Rinri Therapeutics working on hearing loss treatments, and the testimonials sound great. But scaling this in the hyper-regulated, validation-obsessed world of drug manufacturing is a whole other challenge. The potential, however, is undeniable. If you can reliably pick the winning cells early and trash the losers, you compress timelines, improve success rates, and yes, save a ton of money. The CEO’s claim of 80% cost reduction in some steps is staggering. Even if the real-world number ends up being half that, it changes the entire financial model for these therapies. The goal isn’t just cheaper drugs—it’s making transformative treatments that exist today actually accessible. And that’s a future worth predicting.
