Ask any single cell researcher for their biggest experimental bottlenecks, and they will say data analysis. Going back to the beginning of the field, there has always been a “last mile problem,” notes Michael Schnall-Levin, PhD, CTO of 10x Genomics. Why? Because researchers have unique experimental methods and questions, each of which involves some level of writing their own analysis code. It has been “not scalable to write software that can do everything for everyone.”

But now, with advances in LLMs, researchers can use tools to answer questions without needing to be bioinformaticians. Today, 10x is launching Sentira Single Cell, a cloud-based agentic analysis powered by Anthropic’s Claude models as an application for Chromium users.

Schnall-Levin thinks that this will be “massively accelerating” for users. How much faster? “What might have taken someone weeks before could take hours,” he says. And that speed will not only help individual researchers but ultimately, he notes, be beneficial to the pace of scientific research and progress. It is not just about speed, however. Schnall-Levin says that there will like be other benefits. For example, it may reveal results in data that people are not currently seeing.  In addition, the company hopes that Sentira will broaden the group of people who can ask questions of single-cell data directly.

Sentira is, for the company, an umbrella of software capabilities. Sentira spatial was launched a few weeks ago; today’s announcement is specific to single-cell analysis. Sentira is not locked to any model: it can use different LLMs but the Anthropic models are working well, Schnall-Levin reports. And the project follows with Anthropic’s general push into the life science industry.

From virtual lab to AI-powered single-cell analysis

The idea for the Sentira project started at the beginning of 2026, with some work 10x Genomics was doing together with James Zou, PhD, associate professor, Department of Biomedical Data Science at Stanford University. Zou’s name has become infamous with the concept of the “Virtual Lab” which he explains as a laboratory made of algorithms. The idea was first introduced in the 2025 Nature paper, “The Virtual Lab of AI agents designs new SARS-CoV-2 nanobodies.”

Built on 10x Cloud, Sentira Single Cell provides the computing power needed to analyze Chromium data and run multiple analyses in parallel without requiring researchers to configure or manage their own infrastructure. The customer will interact with a chat interface which will help them go through and dynamically ask questions in the data that involves (on the back end) scripting and writing different analysis code.

10x is not training its own LLM model. Rather, it relies on models from Claude. The company will use researchers’ data to understand how the application is performing and to make improvements. But there is no flow back into the models themselves, the company asserts.

The extra cost of going agentic

The company reports that the cost will be an added “modest” cost. More specifically, the company notes that Sentira pricing is credit-based, through 10x Cloud Credit packs. The rough estimate is that each agentic session costs ~12 credits, with complete analysis through to experimental conclusions requiring around 60 credits. At launch, all users will receive 100 promotional credits for the first Chromium dataset uploaded to Sentira Single Cell, enough for users to become familiar with the platform and get to experimental conclusions. Credit packs are available in a range of sizes, with volume discounts that lower the price per credit as pack size increases. (See below for more specific pricing information.)

The promise and peril of AI-driven analysis

When asked about potential concerns about using AI in this way, Schnall-Levin notes (chuckling) that generally speaking he is “far more nervous about AI risks outside of this.”

One point that they discuss within the company, however, is how—as these tools get more powerful—people may have a tendency to rely on them without thinking. And although 10x does believe that the Sentira tools are accurate, he asserts that they are not perfect. “We don’t want someone to just start putting stuff out in a publication with no additional review. We think that this is a really good way to generate initial analysis and exploration, but people are still on the hook and responsible for it.” People need to understand the power and also the limitations, he notes.

Social media is ripe with discussion on the role of AI in single-cell analysis. Luciano Martelotto, PhD, director of global market development for single cell at Waters Biosciences, has expressed opinions—not about Sentira, specifically—but about using AI to analyze single cell data, in general. He notes one worry which is that “nothing in an AI-generated analysis looks broken. The code runs, the plots are pretty, the story makes sense. The mistakes that matter are the silent ones.”

None of the mistakes are “AI being dumb,” he continues. “It does what it’s asked. What it doesn’t do is get that uneasy feeling when a result looks too good. That feeling comes from having watched data go wrong, and I don’t think it can be prompted or coded into existence… not yet, at least.”

And when data analysis becomes increasingly automated and outsourced, will researchers lose the ability to analyze single cell data on their own?  Schnall-Levin sees this as an age old question about technological progress. He points to how few people today can navigate our location by the stars, because modern technology means that we do not need to. But it is a double-edge sword: expertise may be inevitably be lost as people are empowered to accomplish tasks without having to learn the details. Still, the benefits of these new technologies are very likely to outweigh the downside. Either way, this is the direction the world is moving in, and now the single cell field has another tool to move with it.

Pricing Information: