Subscribe to 499
Share Dialog
Share Dialog
Date: December 5, 2025
Event: 499 x HETU | AITOPIA [NO.002]
Topic: AI Agents x Biomedicine: Powering The Next Breakthrough in Healthcare
Over the past decade, AI has transformed many aspects of biomedical research—from predictive modeling to diagnostics. However, we are now standing at a new critical juncture: a leap from passive "AI Tools" to autonomous "AI Agents".
When AI Agents begin to intervene in drug discovery, clinical decision-making, and longevity science, this represents more than just a technical iteration; it is a profound interrogation of traditional medical ethics, data sovereignty, and trust mechanisms.
On December 5, 2025, 499 and HETU co-hosted the second session of the AITOPIA series, titled "AI Agents x Biomedicine." This conversation gathered experts from Avinasi Labs, GenPulse, dLife, GenoSight, and seasoned medical research teams to explore how AI Agents can land in biomedicine through blockchain technology and the boundaries of trust between human doctors and digital intelligence.
Traditional medical AI has often acted as a "single-task executor," such as identifying a specific feature in medical imaging. However, during this seminar, speakers highlighted that the core revolution of AI Agents lies in their Orchestration and Autonomy.
Elaine, Co-founder of GenoSight, pointed out that current AI Agents essentially play the role of a "junior researcher". Unlike traditional AI which performs single-step predictions, Agents can chain models, tools, and datasets into end-to-end reasoning systems.
"In pre-clinical research, agents can already automate literature synthesis, hypothesis generation, and data quality triage. They are not just algorithms; they are workflow orchestrators capable of understanding context and verifying outputs."
Dr. L. A. Wiyadharma, Lead of the Medical Research Team at Celi A1C, added that this autonomy means AI is becoming proactive—capable of initiating and sequencing actions rather than just reacting to user triggers. While this autonomy is strictly regulated in clinical decision-making, Agents are poised to revolutionize administrative utility, such as acting as ambient scribes for SOAP notes and managing patient triage.
Drug discovery has long been trapped by high costs and long timelines—often taking up to 10 years. Winnie Qiu, Co-founder of Avinasi Labs, shared a compelling comparison regarding efficiency:
Traditional Path: Scientists typically take about 18 months to read all relevant papers for a specific condition (e.g., hypertension) and form a hypothesis.
Agent Path: By using AI agent systems to automate reading and hypothesis generation, this process can be compressed to just 3 months.
Winnie mentioned that Avinasi Labs is building an AI-native infrastructure for longevity, helping to tokenize data, models, and IPs so that ownership is decentralized.
In the consumer health sector, Fiona, Founder & CEO of GenPulse, demonstrated the explosive power of combining AI Agents with decentralized data collection. Using female hair loss as an example, she noted that the last drug discovered for this issue was in 1987.
"When training models with centralized hospitals, it took us two years to collect 20,000 images. But after launching our product, we gained over 420,000 unique images from 32 countries. This 40x speed increase in model iteration shows the scalability of going beyond traditional institutions."
Despite the grand technical vision, the core barrier to adoption is not computing power, but Trust and Liability.
When asked, "Do doctors trust recommendations made by AI agents?" the speakers' answers were calm and unanimous: Not yet.
Winnie stated that just as we cannot 100% trust ChatGPT due to hallucinations, we must be even more skeptical in medicine where lives are at stake. Only results validated by rigorous clinical studies hold true value.
Dr. Wiyadharma introduced a profound concept from the clinician's perspective: the "Liability Sink."
"In the current legal framework, the physician is the 'lightning rod.' If the AI hallucinates a diagnosis, the malpractice suit goes to the doctor, not the developer. Until liability can be shared or defined via regulation, the healthcare system will remain paralyzed by risk aversion."
Elaine believes this is exactly where Blockchain plays a role. By settling agent behaviors and reasoning processes on-chain, we can make AI results provable and create an audit trail, gradually building the necessary trust for clinical adoption.
The conversation concluded by returning to the core spirit of Web3—ownership.
Fiona emphasized that the current healthcare system has a massive value disconnect: patients generate valuable data but receive no compensation, while pharma companies spend billions on R&D without direct access to continuous patient data.
The future vision is a DeSci (Decentralized Science) ecosystem driven by Agentic Infrastructure:
Data Assetization: Users generate data via diagnostics or wearables.
Incentivization: Patients are rewarded with tokens for contributing their data to research.
Value Flow: Pharma companies license these validated datasets or models, creating a natural data growth flywheel.
As the host Vincent summarized, the application of AI Agents in biomedicine is no longer a question of whether the technology works, but whether society dares to trust it.
From Avinasi's vision of decentralized longevity, to GenPulse's women's health data network, and GenoSight'sscientific graph generation, we see a new force rising. This force attempts to weave a new web of life with code—where data belongs to individuals, compute serves biology, and Agents become the ultimate bridge between silicon and carbon.
Host: 499
Co-Host: HETU
Special Thanks to Speakers: Winnie Qiu (Avinasi Labs), Walker Chen (dLife), Fiona (GenPulse), Elaine (GenoSight), Dr. L. A. Wiyadharma (Celi A1C).
(For more cutting-edge discussions on AI and BioMedicine, please follow the 499 community and the speakers' projects.)
Date: December 5, 2025
Event: 499 x HETU | AITOPIA [NO.002]
Topic: AI Agents x Biomedicine: Powering The Next Breakthrough in Healthcare
Over the past decade, AI has transformed many aspects of biomedical research—from predictive modeling to diagnostics. However, we are now standing at a new critical juncture: a leap from passive "AI Tools" to autonomous "AI Agents".
When AI Agents begin to intervene in drug discovery, clinical decision-making, and longevity science, this represents more than just a technical iteration; it is a profound interrogation of traditional medical ethics, data sovereignty, and trust mechanisms.
On December 5, 2025, 499 and HETU co-hosted the second session of the AITOPIA series, titled "AI Agents x Biomedicine." This conversation gathered experts from Avinasi Labs, GenPulse, dLife, GenoSight, and seasoned medical research teams to explore how AI Agents can land in biomedicine through blockchain technology and the boundaries of trust between human doctors and digital intelligence.
Traditional medical AI has often acted as a "single-task executor," such as identifying a specific feature in medical imaging. However, during this seminar, speakers highlighted that the core revolution of AI Agents lies in their Orchestration and Autonomy.
Elaine, Co-founder of GenoSight, pointed out that current AI Agents essentially play the role of a "junior researcher". Unlike traditional AI which performs single-step predictions, Agents can chain models, tools, and datasets into end-to-end reasoning systems.
"In pre-clinical research, agents can already automate literature synthesis, hypothesis generation, and data quality triage. They are not just algorithms; they are workflow orchestrators capable of understanding context and verifying outputs."
Dr. L. A. Wiyadharma, Lead of the Medical Research Team at Celi A1C, added that this autonomy means AI is becoming proactive—capable of initiating and sequencing actions rather than just reacting to user triggers. While this autonomy is strictly regulated in clinical decision-making, Agents are poised to revolutionize administrative utility, such as acting as ambient scribes for SOAP notes and managing patient triage.
Drug discovery has long been trapped by high costs and long timelines—often taking up to 10 years. Winnie Qiu, Co-founder of Avinasi Labs, shared a compelling comparison regarding efficiency:
Traditional Path: Scientists typically take about 18 months to read all relevant papers for a specific condition (e.g., hypertension) and form a hypothesis.
Agent Path: By using AI agent systems to automate reading and hypothesis generation, this process can be compressed to just 3 months.
Winnie mentioned that Avinasi Labs is building an AI-native infrastructure for longevity, helping to tokenize data, models, and IPs so that ownership is decentralized.
In the consumer health sector, Fiona, Founder & CEO of GenPulse, demonstrated the explosive power of combining AI Agents with decentralized data collection. Using female hair loss as an example, she noted that the last drug discovered for this issue was in 1987.
"When training models with centralized hospitals, it took us two years to collect 20,000 images. But after launching our product, we gained over 420,000 unique images from 32 countries. This 40x speed increase in model iteration shows the scalability of going beyond traditional institutions."
Despite the grand technical vision, the core barrier to adoption is not computing power, but Trust and Liability.
When asked, "Do doctors trust recommendations made by AI agents?" the speakers' answers were calm and unanimous: Not yet.
Winnie stated that just as we cannot 100% trust ChatGPT due to hallucinations, we must be even more skeptical in medicine where lives are at stake. Only results validated by rigorous clinical studies hold true value.
Dr. Wiyadharma introduced a profound concept from the clinician's perspective: the "Liability Sink."
"In the current legal framework, the physician is the 'lightning rod.' If the AI hallucinates a diagnosis, the malpractice suit goes to the doctor, not the developer. Until liability can be shared or defined via regulation, the healthcare system will remain paralyzed by risk aversion."
Elaine believes this is exactly where Blockchain plays a role. By settling agent behaviors and reasoning processes on-chain, we can make AI results provable and create an audit trail, gradually building the necessary trust for clinical adoption.
The conversation concluded by returning to the core spirit of Web3—ownership.
Fiona emphasized that the current healthcare system has a massive value disconnect: patients generate valuable data but receive no compensation, while pharma companies spend billions on R&D without direct access to continuous patient data.
The future vision is a DeSci (Decentralized Science) ecosystem driven by Agentic Infrastructure:
Data Assetization: Users generate data via diagnostics or wearables.
Incentivization: Patients are rewarded with tokens for contributing their data to research.
Value Flow: Pharma companies license these validated datasets or models, creating a natural data growth flywheel.
As the host Vincent summarized, the application of AI Agents in biomedicine is no longer a question of whether the technology works, but whether society dares to trust it.
From Avinasi's vision of decentralized longevity, to GenPulse's women's health data network, and GenoSight'sscientific graph generation, we see a new force rising. This force attempts to weave a new web of life with code—where data belongs to individuals, compute serves biology, and Agents become the ultimate bridge between silicon and carbon.
Host: 499
Co-Host: HETU
Special Thanks to Speakers: Winnie Qiu (Avinasi Labs), Walker Chen (dLife), Fiona (GenPulse), Elaine (GenoSight), Dr. L. A. Wiyadharma (Celi A1C).
(For more cutting-edge discussions on AI and BioMedicine, please follow the 499 community and the speakers' projects.)
499
499
<100 subscribers
<100 subscribers
No comments yet