Korean Brain-Health AI Startup Voinosis Wins Seoul Pitch for TechCrunch's Battlefield 200

Voinosis, a startup that screens brain health through voice analysis with a claimed 88% accuracy, beat eight rivals at DETO 2026 in Seoul to earn a shot at TechCrunch's Startup Battlefield 200 at Disrupt 2027.

The Seoul conference served as a qualifying round. Hundreds of companies worldwide apply each year to Startup Battlefield 200, the competition that puts early-stage founders in front of venture capitalists and the international tech community at TechCrunch's Disrupt conference. This week, nine South Korean startups pitched at DETO 2026 for one spot in next year's cohort. All nine work in deep tech — healthcare, fintech, energy, semiconductors and AI — and all nine want to enter the U.S. market.

Voinosis won.

How does a five-minute voice check flag cognitive decline?

Voinosis builds an AI tool that analyzes speech patterns — pitch, pauses, rhythm and response times — to detect changes linked to brain health. Each voice check takes less than five minutes.

The company says it has tested the technology in Greek, Japanese and English, and the product does not require separate training for each language. In a clinical study with 146 participants, the startup claims the tech reached 88% accuracy in screening for potential cognitive issues.

Grace Shin founded Voinosis after more than 25 years as an otologic surgeon. The startup targets hospitals, senior living facilities, insurers and individual consumers, and plans to connect voice checks with cognitive training, assessments and treatment from healthcare professionals. Voinosis is raising a Series A round and has already secured a South Korean pharmaceutical company as a strategic investor.

What did the other eight startups build?

Awesome Lab, based in Seoul, makes ThermoFlux, a heating system that generates heat through water electrolysis and ionization. The system manages heat for buildings, water supply and industrial processes, and the company says it cuts energy consumption and carbon emissions. Its U.S. entry plan is specific: supply core OEM modules and develop products with boiler, water heater and HVAC companies.

Crosshub helps travelers and international visitors verify identity and set up payments abroad. Its platform, IDBlock, combines identity verification and payment services using Self-Sovereign Identity and zero-knowledge proof. The company says users can prove identity across borders without sharing unnecessary personal information, and it is integrating the technology with payment systems to widen access to financial services.

DeepFusion AI develops a 4D radar-based imaging system with added sensors for autonomous mobility, robotaxis, maritime surveillance and defense. The system combines RF sensors with radar and lidar to detect objects, movement or distance without a direct line of sight, with all sensors packed into a single customizable unit.

EFLOW builds special motors that produce more torque in less space for robots, drones, satellites and other mobility applications, plus actuators that control how robotic joints move.

Otiton Medical makes two medical devices: one measures body temperature, the other captures ear canal images and analyzes the eardrum. The company says its AI was trained on about 55,000 eardrum images, and the device has received medical certification in South Korea.

Riibotics says it can deploy robots into warehouses within a day. It aims to cut deployment costs by streamlining setup — infrastructure, warehouse mapping, system integration through custom middleware, task programming and safety checks.

Testbank, a Korean edtech startup, is developing Solve, an app that converts printed study materials such as workbooks and test-prep books into interactive digital versions. Its AI identifies question and answer spaces in publishers' PDF files while preserving the original layout, letting users write answers directly on the page and receive automatic grading.

Why does a semiconductor startup from Ulsan matter for AI power bills?

Ternell, the eighth finalist, points to the AI industry's electricity problem. The Ulsan-based company, founded in 2019 by KyungRok Kim, a former Samsung Electronics engineer, is developing a new type of memory to help AI chips use less power.

Unlike conventional two-state memory, Ternell's T-SRAM adds a third state for more efficient data storage. The company reports that tests with 28-nanometer chips showed a 33% to 102% increase in memory capacity per area and a 55% to 71% reduction in the energy needed to access data, compared with conventional SRAM.

What happens next for the winners?

Voinosis advances toward the Startup Battlefield 200 at Disrupt, where it would pitch to venture capitalists and the international tech community. The company's Series A raise, its pharmaceutical strategic investor and its clinical data give it concrete assets for those conversations, and the eight runners-up leave Seoul with a U.S. expansion agenda still on the table.