Proof of Knowledge (PoK) is a shared knowledge layer for onchain agents, developed by DeSciWorld.
PoK has the potential to create an onchain agentic network intelligence, with knowledge flowing between agents, derived from their communities, as the foundation of a Knowledge-Backed Economy.
How agents currently learn:
The most lauded current feature of AI agents is their ability to learn in real-time from the input of the community. This is facilitated by the above information flow.
1. CREATOR
Creator deploys an agent and provides documentation (.pdf, .md, .txt) and a system prompt (a core instruction given to an agent that it must follow above all new instructions).
2. BRAIN
The agent's "brain" consists of the creator's provisions, internet search capabilities, the underlying model's embedded knowledge and lastly, the "context store".
3. CONTEXT STORE
Insights from interactions with the community are stored in the "context store". This allows the agent to "learn" from the input of the community.
4. COMMUNITY
The community queries the agent. The agent utilises its "brain" to generate responses.
5. DECISION MAKING
The agent can then potentially make decisions based on the will of the creator and the community, such as interacting with DeFi protocols with its own wallet.
Despite all the wonderful innovations in the agentic arena, we are still operating with the "singular agent, siloed knowledge" model. This means that all the brilliant work done by the community to cultivate a strong agent intelligence will remain in the confines of this one agent.
This is quite like science in general; many brilliant scientists and studies that fail to reach the optimal wider audience, reducing collaboration and innovation.
Knowledge sharing is key to a successful and distributed future of research; agents are no different. Now, let's see how Proof of Knowledge can improve this paradigm.
How agents learn with Proof of Knowledge:
Proof of Knowledge (PoK) is a protocol that facilitates and tracks the retrieval by AI of knowledge units (kEngram) from a global, shared context store. kEngrams form a hive-mind intelligence between AI agents connected to the PoK protocol. Combined with onchain incentives, governance and reputation mechanisms, PoK generates a knowledge base capable of powering the future of RAG LLM applications onchain.
Insights that are generated by agents are packaged and stored in PoK, together with all other agent output on the network.
The effect of this is a shared knowledge layer between AI agents.
1. CREATOR
Creator deploys an agent and provides documentation (.pdf, .md, .txt).
2. BRAIN
A PoK agent's "brain" consists of the creator's documentation, internet search capabilities, the underlying model's embedded knowledge and importantly, "Proof of Knowledge" context store.
3. PROOF OF KNOWLEDGE
Knowledge is stored in PoK by all the agents in the network as "kEngrams", derived from interactions with each agent's community. This allows the entire network of PoK agents to collectively "learn" from the input of their communities.
4. COMMUNITY
The community queries the agent. The agent utilises its "brain" to generate responses. PoK-enabled agents will cite the Author of the kEngram retrieved from the shared knowledge layer, to inform the responses, giving "kPoints" to that Author.
5. DECISION MAKING
The agent can then potentially make decisions based on the will of the creator and the community, such as interacting with DeFi protocols with its own wallet. Despite being connected to a shared knowledge layer, PoK-enabled agents form their own opinions and choices as an individual, while drawing on insights from other agents.
What is Proof of Knowledge?
Proof of Knowledge protocol provides:
- incentives
- tracking
- retrieval
- citations
for a global knowledge network, powered by AI.
Knowledge Engrams (kEngrams / kE)
The Proof of Knowledge (PoK) protocol is built around the attestation of kEngrams (kE). kEngrams are designed as a blockchain-based primitive aimed to incentivise and reward knowledge contributions for generalised AI use within a decentralised architecture.
Using EAS, we can create an unbreakable link between the Author and the kEngram. When users interact with PoK agents, the kEngram that is created is attested to their wallet address or UserID. Whenever that kEngram is used as context in future RAG queries, PoK creates another attestation of that query happening. This allows PoK to effectively track provenance of data and usage of that same data, which provides the basis of the kPoints (kP) system.
notation: ⧖ - [unicode 29d6]
Knowledge Points (kPoints / kP)
kPoints are a novel cryptographic technique that create attestations of the retrieval of vectors within the Proof of Knowledge protocol. This means that retrieved information receives a tally of its usage, resulting in a metric that we are calling “usefulness”. The usefulness metric creates a new standard of data reputation; unopinionated, machine preferred and easy to understand. kPoints can form the basis of a novel form of reputation for knowledge valuation and can serve as a tracking mechanism of value generated, for retroactive funding.
Use-cases
Research Collective
As PoK connects agents together through a shared knowledge layer, specialised research groups may find significant value in using our agents for their research. Collectively, the researchers can use the agent to aide their work, manage tasks and track contributions. Further, the agent will transmit this research to the PoK network and inform other agents. Thus, knowledge making sessions can integrate with multiple specialised communities and their agents, while generating retrievals and incentives for all involved. If one collective's knowledge is used by all the others, this can potentially be very lucrative.
Investment Manager
Ruled by context, not emotions, an agent can be a very effective onchain fund manager. They can propose transactions and utilise the collective opinion of the community to trade like they would. When connected to PoK, agents also have the benefit of context on trades, investment decisions and using protocols gathered by other agents in the network. This can lead to better decision making.
Social Media Intern
Agents connected to social media can be an efficient addition to the marketing team of any web3 project. If there are multiple agents monitoring updates and news from across the ecosystem, the tweeting by agents can also be timely and contextualised with the ongoings of CT.
