Unlocking liquid science markets through tokenized IP
Joined April 2018
- Tweets6.8K
- Following2K
- Followers30.3K
- Likes9.5K
Molecule retweeted
Partnering & licensing involve letting someone outside your org read confidential work.
With Molecule Labs you can grant a partner timeboxed, read-only access and files are encrypted before they leave your device. There's no need to take a security trade off to collaborate.
If you're interested in implementing Labs in your pipeline, book a call.
molecule.xyz/solutions/secur…
Molecule retweeted
What's new in Molecule Labs?
→ Lab owners & contributors can now edit a file's details, switch it between public and confidential, or delete it, from a menu on each file
→ The Developers page lets you create, view and revoke your own API credentials, completely self-serve
ALT Edit, delete or change visibility of files
What's new in Molecule Labs?
→ Lab owners & contributors can now edit a file's details, switch it between public and confidential, or delete it, from a menu on each file
→ The Developers page lets you create, view and revoke your own API credentials, completely self-serve
ALT Edit, delete or change visibility of files
Molecule retweeted
In a new paper, researchers at @GoogleDeepMind ask what science needs once AI can come up with ideas faster than labs can test them. Their answer includes a fair way to choose which ideas get tested, and a record of who contributed what, whether a person or an AI agent.
Molecule's protocol architect Phill Lee (@0xPho_) went through their proposal, mapping out where Molecule Labs has a role to play.
In a new paper, researchers at @GoogleDeepMind ask what science needs once AI can come up with ideas faster than labs can test them. Their answer includes a fair way to choose which ideas get tested, and a record of who contributed what, whether a person or an AI agent.
Molecule's protocol architect Phill Lee (@0xPho_) went through their proposal, mapping out where Molecule Labs has a role to play.
Today a Lab can:
• give an AI agent or a partner lab access to a project's data for a set period, and revoke it at any time
• record every contribution with a timestamp and its author
• keep sensitive data confidential while sharing the rest publicly
Payouts when results are validated, and revenue split by contribution, are next on Molecule's roadmap.
Phill's full piece takes each DeepMind proposal in turn: molecule.xyz/blog/onchain-la…
Researchers at @GoogleDeepMind, expect that as routine research is automated, the human scientist becomes "the principal, underwriting the actions of AI delegates". That depends on the scientist controlling what each agent may access and seeing what it did.
A Molecule Lab lets one researcher give each AI agent access to its data for a set period, revocable at any point, with every grant and every contribution logged against its author, including AI agents. PeptAI (@peptai_) already works this way, with an AI agent running peptide research under a human lead and reporting through its Lab.
Full blog by Molecule's protocol architect Phill Lee (@0xPho_) molecule.xyz/blog/onchain-la…
As AI pushes the cost of generating a scientific hypothesis toward zero, the bottleneck moves to validation, because trial patients and laboratory time stay scarce.
@weballergy and colleagues at @GoogleDeepMind treat this as a market design problem in "Agentic Economies for Autonomous Scientific Discovery", and their proposal asks a lot of its onchain layer. The central research asset would need to:
> anchor ideas to a tamper-evident record gate
> price access to data hold funds
> release them on validation split credit along a provenance graph
> support community ownership
> give AI agents bounded permissions that can be revoked
@0xPho_, Molecule's protocol architect, has published a response mapping these requirements onto the Onchain Lab, a programmable smart account bound to an NFT whose holder controls the Lab. Each Lab links to an encrypted data room where human and AI contributors hold expiring roles, and attested modules extend what it can do without migrating assets.
Molecule designed the Onchain Lab in 2025 and deployed it in June 2026. Expiring, onchain-logged roles for agents and pay-per-query data access run on Labs today, while royalty escrow and revenue distribution are on the roadmap as modules that reach every existing Lab.
Read @0xPho_'s full response:
molecule.xyz/blog/onchain-la…
A new 'Help' menu brings What's New, Docs and FAQ into one place.
The What's New feed collects recent product updates, and a dismissible card appears in the sidebar whenever something new ships, so users can stay up to date with all product improvements.
We've improved our Set Up guide to help users explore the full functionality of the lab from the start.
Lastly, Labs now sends a welcome email when someone creates their first Lab, and emails a member whenever their role on a Lab changes.
All live at app.molecule.xyz
We've improved our Set Up guide to help users explore the full functionality of the lab from the start.
Lastly, Labs now sends a welcome email when someone creates their first Lab, and emails a member whenever their role on a Lab changes.
All live at app.molecule.xyz
Molecule's initial intellectual property primitive was highlighted in a paper from @GoogleDeepMind's research team: "Agentic Economies for Autonomous Scientific Discovery"
We believe the future sits at the intersection of scientific, technological, and legal innovation.
Happy to share a new preprint: "Agentic Economies for Autonomous Scientific Discovery"
arxiv.org/abs/2609.31562
that encapsulates our thoughts on the AI revolution in science, and how applying an economic lens enables us to better recognize the key bottlenecks and how to address them.
For those of you curious about whats happened in the four years since the IP-NFT was launched - our recent protocol whitepaper covers our development. We'd love to know what you think.
molecule.xyz/blog/molecule-p…
Our thanks to @weballergy, @FranklinMatija, @sebkrier, @sindero, @Dr_Atoosa, @vivnat & @alan_karthi for sharing your insights on the AI revolution in science.
Full paper is linked below arxiv.org/abs/2609.31562