[đď¸On the Record #7] Meet Portrai, a Seoul startup mapping the human body for drug discovery
Most drugs never make it to patients. These four Seoul doctors are trying to fix that.
Every year, the pharmaceutical industry spends billions searching for cures. Most of it vanishes. According to McKinsey, the attrition-adjusted cost to develop a single drug asset is now estimated at up to $2.8 billion, and still, only about 13% of assets that enter Phase 1 trials ever make it to launch.
Not because the science was wrong. But because no one knew where to look. Imagine navigating a city with no map. No streets, no addresses â just a list of buildings. Thatâs been drug development for the last 30 years.
Portrai is building that map. In 2021, four physicians from Seoul National University College of Medicine, tired of watching good science fail for the wrong reasons, set out to find out why.
The human body is not a spreadsheet. For decades, drug researchers have treated it like one. Bulk sequencing, the dominant method of genomic analysis, grinds tissue for analysis, extracts the data, and calls it insight.
But as Portrai CEO Daeseung Lee put it: âThis is akin to looking at a blended fruit smoothie and trying to guess the original arrangement of the fruits.â
With solid tumors, that missing context is critical. Traditional approaches canât capture the complex interactions between cancer cells and the Tumor Microenvironment (TME). You learn what genes are active. You lose everything else: the architecture, the neighborhoods, the conversations happening between cells.
That gap, it turns out, is where drugs go to die. When a promising compound enters a clinical trial, it isnât navigating a flat dataset. Itâs navigating a three-dimensional ecosystem, and without a map, neither the drug nor the researchers who designed it know where to go.
Their answer was to build one. Lee brought IT scaling expertise. CMO Kwonjoong Na brought oncology. CTO Hongyoon Choi and CSO Hyungjun Im were the ones who saw the mathematical solution, if spatial context was the missing variable, then spatial transcriptomics combined with AI was the answer. Together, the four made the case for what they now call the âlocationâ problem in drug delivery.
The result is a platform that maps RNA activity at the cellular level, spanning oncology, nuclear medicine, nanotechnology, and IT.
Spatial transcriptomics, a technology that reads RNA activity cell by cell while preserving each cellâs exact physical location within tissue, is not a new concept. But doing it at scale, with clinical-grade precision, on real-world patient data, is where most players stop.
Portrai says it has gone further. The company has built what it calls a Human Tissue Atlas, claimed to be the worldâs largest spatial transcriptomics database, covering both tumor microenvironments and normal tissues. On top of that data sits its proprietary Vision-Gene-Language (VGL) model, a multimodal AI that reads histology images, gene expression patterns, and clinical language simultaneously.
Instead of asking âDoes this drug work?â, Portrai can ask âWhere does it go, what does it hit, and what does it miss?â
For next-generation cancer therapies, ADCs (Antibody-Drug Conjugates) and RPTs (Radiopharmaceutical Therapies) in particular, this distinction is everything. These are precision weapons. Precision weapons need precise maps.
The company has completed over 50 client projects and received acceleration support from Amgen, J&J, and NVIDIA. In late 2025, it closed a deal with Celltrion, a collaboration to co-discover novel drug targets using PortraiTARGET, with a total contract value of up to $87.75 million, including milestone payments and royalties.
Also in late 2025, Portrai closed a KRW 14.5 billion (approximately $9.6 million) Series B, backed by Stic Ventures and Kakao Ventures, among others, a signal that Korean venture capital is paying attention.
What comes next
Portrai is now in conversations with big tech companies. Details remain undisclosed. But the direction is clear: the map is attracting attention beyond pharma.
The five-year plan goes beyond oncology. Portrai is working on tools to simulate how drugs behave in patients before they ever enter a trial, and building out a pipeline of its own cancer drug candidates in ADCs and RPTs. If the platform works the way the founders say it does, the eventual targets are neurodegenerative diseases and inflammatory conditions too.
For now, the startup has 26 employees across offices in Seoul and Boston. The map is still being drawn.
The full transcript of this interview is available on Patreon. If you're into in-depth founder conversations, that's where I go deeper.

