"How" Part 1 - Data at Scale
Big vision will be fulfilled by small actions day-by-day.
Avinasi Labs: How We Are Solving Longevity’s Biggest Problems
In our earlier piece, we identified four major problems slowing down progress in longevity:
Scientific limitation – We still don’t fully understand the causal mechanisms of why we age.
Linearity of development – Biology and clinical development remain linear. Traditional trials do not scale for longevity interventions.
Infrastructure gap – Except for Singapore, healthcare systems globally remain focused on “sick care,” not preventative care. Incentives and regulations are designed to treat disease, not to preserve health and longevity.
Equitable longevity – Access to both health and longevity remains deeply unequal across the world.
We strongly believe that without execution, an idea is worthless, and a dream is only a dream. So, how are we going to solve these problems?
Step 1: Building the Foundation – Data at Scale
The first problem: our limited understanding of aging, cannot be solved without large-scale, diverse, high-quality data. That is why the cornerstone of Avinasi Labs is building an ecosystem that can collect and generate data at scale.
There are two primary ways to do this:
Products that generate data – Wearables, blood testing platforms, and even robotics that produce biological or behavioral data (e.g., Apple Watch, Whoop, Oura, or next-generation humanoid robots). Apps and use cases built on top of these devices also fall into this category.
Products that require data as input – Tools where the user must input personal data to generate valuable insights. Think of ChatGPT for biology: the model cannot give you meaningful outputs without understanding your data.
By combining both approaches, we can generate the datasets needed to tackle the root question of aging: why do we age, and how can we prevent it?
Step 2: The Avinasi Approach – A Foundation Model for Longevity
At Avinasi Labs, our approach has clear founder–product fit. We are building a foundation model for longevity, composed of three layers:
1. The Foundation Layer
This is where my co-founder, Dr. Albert Ying, brings unique expertise at the intersection of longevity biology and computer science.
We are training a vertical foundation model for longevity by:
Integrating all publicly available datasets.
Fine-tuning with domain-specific architecture.
Establishing new heads and updated neural networks to yield outputs that matter, such as mortality risk prediction, causal inference on biological pathways, and validated interventions.
Technically, this means moving beyond generic LLMs. We are constructing domain-specific embeddings for biological data and coupling them with neural architectures optimized for longitudinal prediction and intervention ranking.
2. The AI Agent System
On top of the foundation model, we are building autonomous AI agents that can automate workflows. Two initial systems are in development:
Drug discovery agent (pre-clinical stage): Scientists can upload lab notes or experimental results. The foundation model and agent system will mine biological data, identify relevant mechanisms, propose molecules, validate findings against existing publications, assess trustworthiness, and even generate a draft research summary. For the first time, we are automating the scientist’s desk work end-to-end.
Consumer health agent: Users interact with it like ChatGPT, but instead of text answers, it integrates with their health data (blood tests, wearables, lifestyle inputs). It predicts healthspan and lifespan, orders tests, and optimizes routines. Unlike current “daily readiness scores,” it delivers hour-by-hour predictions and personalized interventions.
3. The Application Layer
Great science is not enough. For consumers, experience is everything. Steve Jobs’ philosophy of product taste is a guiding principle here: intuitive products do not force behavior; they enhance what users already do, and surprise them with insights they didn’t expect.
In longevity, this means:
Universal apps for common needs like sleep optimization.
Specialized apps for more scattered needs, like posture correction or fitness.
After interviewing longevity enthusiasts, one theme stood out: energy. As people age, energy is the first thing they feel slipping away, and they have already tried supplements, exercise, and sleep hacks to get it back.
That’s why our flagship product is Sponge, an app that predicts and optimizes energy levels, hour by hour. Instead of generic readiness scores, Sponge uses our foundation model to generate real-time recommendations that actually fit into users’ lives.
You’ll be able to download Sponge on October 3, 2025, after its launch at Token2049 in Singapore.
Step 3: Beyond the First Product
Sponge is just the beginning. It is likely the first product in Web3 that combines a real-world use case with AI and longevity, not just financial incentives.
But to solve longevity at scale, we still need to address the other three challenges:
Biology’s speed limits
Infrastructure gaps
Equity in access
These will require systems built both sequentially and in parallel. Some will be addressed through infrastructure, others through ecosystem collaboration. They are outside the scope of this article, but in future updates, we’ll share how we are tackling each of them.
What matters now is this: Avinasi Labs is laying the foundation, literally, for a new era of longevity research and applications.
This is not a small vision. We are building a system that requires coordination and collaboration at scale, but also one that can change the trajectory of human health and lifespan.
Last updated