Prisoner’s Dilemma is the single largest impediment to human coordination.
It's why wars perpetuate longer than they should; it's why perfectly cogent acquisitions never happen; and it's also why putting together a plan with your friends is so hard.
No one wants to be the first to commit. Even attempting to broker a negotiation changes the entire dynamic. And finding consensus each party would collectively benefit from, structurally stymies to exist.
We solve this, and bypass the sequential commitment lock in human negotiation with a new paradigm of multi-state agent negotiation.
We’re not a Neo lab, we don’t do AI research.
Our work gears far more to framing and defining the RIGHT set of problems to tackle in a post-AGI world.
Asking the right questions has always been perniciously harder than finding the answer.
Which is why we make a heavy bet on young talent, and push tirelessly for novel abstractions, without being too quick to define them.
We host an exclusively math centric team, with 20+ IMO Golds, IOI Golds and campers from MIT, Stanford, Cambridge, Harvard and AG.
Among the research we do, we rethink how trust scales; solve human coordination with social contracts; solve Blame Assignment in RL; and now even challenge intellectual property with fractional IP, bartered through agent-agent negotiations.
We are protocol builders.
It takes the most uncanny of ideas to fundamentally change how we look at the world.
And SPL seeks to propound just those.
-Shivang Raisurana
Shivang Raisurana
Built a computer graphics lab with CEO of Amazon India; Was the prime investigator of the world’s largest medical study with WHO and the supreme court of India; National Level Rifle Shooter.
CEO
Aditya Jain
IOI Coach of India; 3rd highest rated competitive programmer in Indian history; 2x ICPC World Finalist. Built the entire quant infra for a 7Billion AUM fund, valued >$100mn in 1 year.
CTO
Mason Eyler
Putnam Top 25 in the USA, Top 2 Math grads at Stanford this year. Endurance Runner; Inventor of the Trust Graph.
COO
Our coding language, Witness
The bedrock protocol that allows agents to interact without any scope of malicious context extraction or prompt injections.
Social Contracts
Programmatic agreements that people enter with each other, though their agents.
The Trust Graph
A way to dynamically imbue relational trust between any species of nodes on a graph.
Mega Nodes
A node that captures the emergent trust state of a collection of nodes.
Paramaterised Nodes
How each node on the graph is governed by its own custom algorithm.
Beat Google Search
When the trust graph became a living breathing organism that calibrates itself.
Private Agent Negotiation
How we use agents as stateless machines with ephemeral memory to coordinate sensitive information.
Solving Prisoners Dilemma
How we engender a series of constrained agent proxies to handle multiple states simultaneously.
The Conditional Pledge
A new paradigm of open sourcing for labs, and researchers to share IP without leaking it.
Abstraction Matrix
A world simulation sandbox for governments and companies to visualize multi-agent workflows.
Solving Blame Assignment in Reinforcement Learning
An alignment technique to solve long term thinking, deep research and self critiquing, by making AI think for its future self.
