Origin: Using AI to Make Gene Therapies Safer

Origin is an early-stage biotechnology startup working at the intersection of artificial intelligence and genetic medicine. Founded in 2025 and based in San Francisco, the company is part of the Winter 2026 startup batch and currently operates with a lean team of four. Despite its small size, Origin is tackling one of the most consequential challenges in modern medicine: how to make cell and gene therapies safer, more precise, and more controllable.

Cell and gene therapies represent one of the most promising classes of treatments developed in recent decades. They have already shown success in conditions once thought untreatable, including advanced cancers, rare genetic disorders, and degenerative diseases. However, the broader adoption of these therapies has been slowed by a critical issue—lack of precise control over gene expression. Origin exists to solve that problem by re-engineering the regulatory DNA that determines when, where, and how strongly therapeutic genes are activated.

By combining AI-driven sequence design with large-scale experimental validation, Origin aims to redefine how regulatory DNA is discovered, tested, and deployed in therapeutic contexts.

Who Are the Founders Behind Origin?

Origin was founded by Yash Rathod and Malhar Bhide, two entrepreneurs with a shared vision of transforming genetic medicine through computational design. Yash Rathod serves as CEO, while Malhar Bhide is the company’s CTO. Together, they lead Origin’s scientific and technical direction, balancing strategic leadership with deep involvement in model development and experimentation.

The founders describe Origin’s mission succinctly: to use AI to design “DNA switches and dials” that can program gene expression with unprecedented precision. This framing captures the core idea behind the company’s work—genes themselves are not enough. What truly determines therapeutic success is the regulatory code that governs gene behavior inside specific cells and tissues.

Their combined expertise spans AI modeling, genomics, and experimental biology, positioning Origin to operate across both computational and wet-lab domains rather than relying solely on in-silico predictions.

Why Are Cell and Gene Therapies Both Powerful and Risky?

Cell and gene therapies have already delivered life-changing outcomes. Patients with stage IV leukemia, inherited retinal diseases, and spinal muscular atrophy have benefited from treatments that directly alter cellular function at a genetic level. Unlike traditional drugs that must be taken continuously, these therapies can offer long-lasting or even curative effects after a single intervention.

However, the same power that makes these therapies effective also introduces risk. Many toxicity events arise because therapeutic genes are expressed in the wrong cells, at the wrong time, or at excessively high levels. Once a gene is introduced, controlling its behavior becomes extremely difficult without precise regulatory mechanisms.

As a result, promising therapies for cancer, autoimmune diseases, neurodegenerative disorders, and even aging are often limited by safety concerns rather than lack of efficacy. Origin’s work directly addresses this bottleneck by focusing on regulatory DNA—the genetic elements responsible for controlling expression patterns.

What Role Does Regulatory DNA Play in Gene Expression?

Regulatory DNA sequences, such as enhancers and promoters, function as the control layer of the genome. While genes encode proteins, regulatory elements determine when those genes are activated, in which cell types, and at what intensity. They interact with transcription factors and chromatin structure to orchestrate complex expression programs across different tissues and disease states.

Despite their importance, regulatory sequences remain poorly understood and difficult to design. Traditional methods rely heavily on known motifs, repeated patterns, or random ligation approaches. These strategies often produce weak candidates or require brute-force screening of massive libraries to identify a single functional sequence.

The lack of systematic design principles for regulatory DNA is a major obstacle in therapeutic development. Origin was founded on the belief that this problem is fundamentally computational—and therefore solvable with the right models and data.

What Is the Core Problem Origin Is Solving?

The central problem Origin addresses is the inability to reliably control gene expression in therapeutic contexts. Current regulatory sequence design approaches are inefficient, imprecise, and poorly suited to the complexity of real biological systems. They fail to account for the diversity of cell states, transcription factor interactions, and chromatin environments that determine whether a therapy succeeds or causes harm.

Because of this, toxicity events often emerge late in development or clinical trials, increasing costs and limiting patient access. Origin’s founders argue that without a new approach to regulatory DNA design, the full promise of cell and gene therapies will remain unrealized.

Their solution reframes regulatory sequences not as static motifs to be copied, but as programmable elements that can be designed, tested, and optimized at scale using AI.

What Is Axis and How Does It Work?

Axis is Origin’s state-of-the-art AI model designed specifically to generate and evaluate regulatory DNA sequences. Unlike traditional predictive models that score existing sequences, Axis is generative—it creates entirely new enhancers and promoters tailored to specific target cell states.

When prompted with a particular disease-relevant cell type, Axis can rapidly sample millions of candidate regulatory sequences. These sequences are designed to activate therapeutic genes selectively in the intended cells while minimizing off-target expression elsewhere.

Axis is trained not only to design sequences but also to screen them, ranking candidates based on predicted transcriptional activity, specificity, and molecular properties. This dual capability allows Origin to explore a vast design space far more efficiently than brute-force experimental methods.

How Is Origin Building Its Proprietary Dataset?

A defining aspect of Origin’s strategy is its commitment to experimental validation. Rather than relying solely on public datasets or computational benchmarks, the company is actively testing Axis-generated sequences in the lab.

These experiments measure how synthetic regulatory elements interact with transcription factors and how they influence phenotypic outcomes of gene expression. Over time, this process is producing what Origin describes as the largest proprietary dataset of synthetic regulatory DNA sequences across diverse cell and tissue types.

The dataset consists of millions of experimentally validated sequences, giving Origin a compounding advantage. Each new experiment improves the model, which in turn generates better candidates, accelerating the cycle of discovery.

How Does Axis Perform Compared to Existing Models?

Origin reports that Axis demonstrates strong performance across multiple predictive benchmarks that measure molecular properties of real regulatory sequences. These include chromatin accessibility and transcription factor binding—key indicators of whether a regulatory element will function effectively in a biological context.

Notably, Axis has been shown to outperform Google DeepMind’s AlphaGenome in predicting regulatory element activity. While AlphaGenome is widely recognized as a leading model in genomic prediction, Origin’s results suggest that models trained specifically on synthetic regulatory design may hold an advantage for therapeutic applications.

This performance is not limited to prediction accuracy. Axis also excels in generating sequences with high diversity, exploring regions of the design space that traditional methods rarely reach.

Why Does Sequence Diversity Matter?

One of the risks in regulatory DNA design is over-reliance on known biological patterns. While familiar motifs can be effective, they also increase the likelihood of unintended interactions and off-target effects.

Axis addresses this by generating highly diverse sequences. According to Origin’s internal analysis, over 72% of Axis-generated sequences have no BLAT alignment match among themselves when using a 20-base-pair threshold. This indicates that the model is not simply remixing known patterns, but discovering novel regulatory logic.

High diversity expands the search space for safe and effective regulatory elements, increasing the likelihood of finding sequences that are both potent and specific.

How Is Axis Independently Validated?

To assess the quality of its generated sequences, Origin used Malinois, an independent and in-vitro validated prediction model developed by the Broad Institute. Using Malinois, the team observed that Axis-generated sequences exhibited the highest predicted transcriptional activity in the cell types they were prompted for.

This external validation strengthens Origin’s claims by demonstrating that Axis performs well not only within its own modeling framework, but also when evaluated by established third-party tools.

What Diseases Could Origin’s Technology Impact?

While Origin’s platform is broadly applicable, its founders highlight several areas where precise gene regulation is especially critical. These include cancer, central nervous system disorders, autoimmune diseases, neurodegenerative conditions, and potentially aging-related pathologies.

In each of these domains, therapeutic success depends on activating genes in highly specific cell populations while avoiding others. Origin’s approach makes it possible to design regulatory DNA tailored to these nuanced requirements rather than relying on one-size-fits-all solutions.

What Is Origin’s Long-Term Vision?

Origin envisions a future in which regulatory DNA design is as programmable and predictable as software. By turning gene expression into an engineerable system, the company aims to remove one of the biggest barriers to safe and scalable genetic medicine.

In the long term, Origin’s growing dataset and modeling infrastructure could become a foundational layer for the next generation of cell and gene therapies. Rather than discovering regulatory elements by trial and error, researchers could design them on demand, with confidence in their safety and efficacy.

Why Could Origin Be a Defining Company in Genetic Medicine?

Origin sits at the convergence of AI, genomics, and experimental biology—three fields that are rapidly reshaping medicine. Its focus on regulatory DNA addresses a fundamental limitation that has constrained cell and gene therapies for years.

By combining generative modeling, large-scale validation, and a proprietary data moat, Origin is building more than a single product. It is creating a platform that could redefine how therapeutic genes are controlled inside the human body.

If successful, Origin’s work could help unlock durable, safer cures for some of the most complex diseases humanity faces, marking a significant step forward in the evolution of precision medicine.