Introduction
Deepcell is a Menlo Park life sciences company spun out of Stanford University in 2017 that applies artificial intelligence to label-free single-cell morphology — capturing, analyzing, and sorting cells based on high-dimensional morphological features rather than fluorescent tags alone. Maddison Masaeli, PhD, serves as co-founder and chief executive officer per Deepcell’s April 21, 2026 press release launching REM-I OnSite, a fully on-premises deployment of its Axon Data Suite software.
Deepcell’s REM-I platform uses what the company calls the Human Foundation Model to identify and sort cells by morphological distinctions, supporting basic and translational research across oncology, cell therapy, immunology, and ophthalmology. The April 2026 OnSite configuration runs run management, analysis, and visualization locally on the instrument with no cloud dependency — keeping research data air-gapped for regulated and data-sensitive environments.
Deepcell positions morphology profiling as complementary to genomics — capturing functional and visual phenotypes that sequencing alone may miss, particularly in cell therapy manufacturing and heterogeneous tumor microenvironment studies where labeling workflows alter cell behavior.
What the company does
Deepcell combines AI, microfluidics, and high-resolution imaging to profile single cells without traditional labeling workflows that can alter cell behavior. The Axon Data Suite manages runs, morphological analysis, and visualization — previously cloud-connected for many customers, now available entirely on-device via REM-I OnSite. Researchers use morphology embeddings to discover subpopulations, quality-check cell therapy products, and explore biological variation invisible to standard flow cytometry panels alone.
Deepcell positions morphology as an orthogonal data modality complementing genomics and proteomics — particularly where labeling is impractical or where visual phenotypes carry functional information AI models can quantify at scale.
Who it serves
Deepcell sells to pharmaceutical R&D teams, academic core facilities, and GxP-regulated laboratories where data governance policies prohibit routing competitive morphology datasets through external cloud infrastructure. The April 2026 release explicitly names large pharmaceutical organizations with stringent data policies and international placements in China and other markets with data localization laws as target buyers for OnSite deployments.
Chief technology officer Mahyar Salek, co-founder, is quoted emphasizing that REM-I generates among the most sensitive and competitively valuable data Deepcell customers produce — making on-device retention a requirement rather than optional feature for many accounts.
Company background
Deepcell originated in Stanford laboratories with co-founders Maddison Masaeli and Mahyar Salek building AI-powered cell analysis tools that bridge life science, data science, and hardware engineering. The about page lists a scientific advisory board including Stanford faculty and Broad Institute researchers, reflecting the company’s academic roots. Deepcell remains headquartered in Menlo Park, California.
Masaeli remains the quoted CEO in product launch communications; Salek leads technology strategy as president and CTO per the about page leadership roster reviewed July 10, 2026.
Product and AI capabilities
Deepcell’s Human Foundation Model learns morphological representations from imaging data at single-cell resolution, enabling classification and sorting decisions without predefined antibody panels. REM-I OnSite extends the same AI stack to environments where network egress is blocked — run management, analysis pipelines, and visualization execute on the instrument with all data remaining local.
For international expansion, OnSite removes a common adoption barrier: customers in regions restricting Western cloud infrastructure can still deploy Deepcell’s full software experience. Deepcell’s commercial team configures deployment models per institution compliance needs rather than forcing a single SaaS architecture.
Key developments
2017: Deepcell spun out of Stanford University; founded in Menlo Park. 2020s: REM-I platform commercialized across pharma and academic research. April 21, 2026: REM-I OnSite launched for fully on-premises, air-gapped Axon software operation on the instrument. April 2026 release targets GxP facilities and international data-sovereignty requirements.
Masaeli and CTO Mahyar Salek are both quoted in the April 2026 release — CEO on customer-required deployment terms, CTO on data never leaving the room where cells are analyzed — reflecting coordinated product and governance messaging for regulated buyers.
Why it matters
Deepcell matters because single-cell research increasingly collides with enterprise data governance — cloud-first SaaS models fail in pharma R&D and certain national jurisdictions even when algorithms perform well. REM-I OnSite tests whether morphology AI vendors can win regulated accounts by matching deployment flexibility, not only model accuracy. Compare with TechBio discovery platforms such as Recursion Pharmaceuticals that generate phenotypic data at different biological scales.
Label-free morphology sorting also addresses cell therapy manufacturing quality control — where fluorescent markers may not reflect functional state Deepcell’s models quantify from imaging alone. Pharma partners quoted in the April 2026 release treat on-device data retention as contractual requirement, not optional IT preference.
Sector context
Deepcell belongs in Brel’s AI in healthcare coverage under life-sciences research AI and single-cell biology, related to drug-discovery TechBio companies such as Recursion Pharmaceuticals and Owkin. Clinical diagnostic AI vendors operate under different regulatory paths than Deepcell’s research-platform deployments.
REM-I OnSite aligns with 2026 pharma IT trends toward sovereign AI infrastructure — morphology datasets that cannot cross national borders require on-device analytics vendors traditional cloud SaaS models cannot serve without architectural rewrites.
Sources and references
Deepcell REM-I OnSite press release (April 21, 2026) and about page with leadership.
- Deepcell — REM-I OnSite launch (deepcell.com, April 21, 2026)
- Deepcell — About (deepcell.com)
Reviewed July 10, 2026. Maddison Masaeli, PhD, is co-founder and CEO. Spun out of Stanford 2017; Menlo Park headquarters.