- banksyweb3
- Posts
- Reppo - the composable intelligence framework
Reppo - the composable intelligence framework
Reppo
Reppo have a Compossible Intelligence Network (CIN), leveraging zkTLS and programmable IP to enable AI developers to become entrepreneurs onchain. Reppo allow DePINs and SMEs to share their metadata (usually proprietary data) which is composed into derived IP. Then, they have a co ownership model to bootstrap DePINs and tokenomics models and AI models to fast track GTM.
Reppo Steps
“Our ultimate vision is to see the world's first active inference foundation model come to life, built collaboratively on purpose-built AI models, agents, and systems that are governed and owned by developers who advance builder-owned intelligence with intention to push humanity many steps closer to sovereign AGI.”
Reppo.xyz
reppo.xyz enables AI specialists to use their data, to create models which are highly tuned to detect gaming of or optimisation of the network. This enables them to generate maximal value out of their DePINs, whilst minimising negative impacts. AI models are being recognised as digital assets, with their creators benefiting from the value in their datasets.
Reppo.ai
reppo.ai is a marketplace to monetise the AI models and IP generated by their creators.
Reppo Timeline
Jan 2024 - Reppo was spun out of Protocol Labs.
H1 2025 Testnet launch.
H2 2025 Mainnet launch.
Reppo Protocol
The Reppo Porotocol is “a standard that enables co-ownership and co-monetization schemes for model developers and enterprises.”
Reppo’s Model Pool is where models are registered by their owners, to set their pricing and IP terms and conditions. Developers can then earn royalties when their AI IP is used. The Model Pool will be built on the Base network for testnet.
ModelRivet - Global State
AI infrastructure is abundant, but fragmented, ModelRivet aims to orchestrate this by allowing users to own and monetise their AI IP onchain. ModelRivet users will upload their data, and set monetisation terms for their data to be listed on the Reppo exchange.
“Reppo pods will use ModelRivet to satisfy end-customer demand.”
Reppo Intelligence Pods - Local State
“Reppo Pods are sovereign Human +AI collectives that act as Inference DAOs.”
Reppo pods are digital assets which are owned by individuals or groups, who then govern and monetise the AI assets and IP in the pod. They are designed to be sovereign. End users can interact with the pods directly or have their requests routed via Reppo’s interface network, using Anoma’s intent solvers. Domain specific inferencing requests will be routed to pods as intents, which are then satisfied by the specialist pod. The pods will serve queries for specific use cases, rather than a generalised LLM. Pods are powered by ModelRivet. Each pod will have reputation mechanics (see below).
Reppo Intelligence Clusters can form, where pods receive token rewards. Also, Pods can be a public good, by sharing their data with other pods and the wider Reppo community.
Reppo Staking
The Modular Reputation Framework
“At Reppo, we are building the world’s largest database and marketplace for risk & reputation assessment models. These models range from AI/ML models that help assess climate change risk to assessment models for Web3 networks.”
The Sovereign Risk & Reputation Intelligence Cluster on Reppo allows teams to understand nodes, network data and their reputation. AI/ML models allow dynamic scoring for the reputation of nodes and users, using metrics such as; network KPIs, user reviews, reliability, geographical decentralisation, and patterns. Patterns and collusion can be detected.
Reppo - The Modular Reputation Framework
DePINs
Reppo aim to automate the self governance layer for DePINs. LLMs can spot patterns of good and bad actors on a network, this becomes like an “immune system”, where good and bad onchain actions have automated responses. Reppo have created DePIN models with DIMO for emissions scores and green modelling. In addition to, Casper Studios’ Pod, for AI Agents to carry out automated Solar Farm audits on the Glow Protocol.
Reppo created Fielcoin’s reputation systems, to give nodes a dynamic score based on operational metrics. A Miner Performance Prediction Model and a Filecoin Allocator Reputation Model were hosted on Open Gradient.
Reppo’s Filecoin models
DeAI
Reppo were spun out of Protocol Labs to take DRL algorithms learning and create a Web3 reputation system. Reppo challenge the notion that you cannot monetise open-source. AI can be used to govern at scale, catch bad actors and automate rewards for positive network contributions. AI can also be used to price risk and inform tokenomics design.
Computational liberty is best achieved via technological, not legal means (credit: Gensyn).
Intents
Reppo use Anoma for intent-sourcing and routing to pods. Reppo are targeting a Chat-GPT like UI/UX.
Reppo Partnerships
Reppo are powered by @anoma and secured by @Gateway_xyz. Reppo developed due dilligence models for Architect.systems (angel investor). Reppo have partnered with Vana Protocol, a ”distributed network for private user-owned data” to enable modelers to receive downstream payments after creation. Vana’s data DAOs will provide further data onboarding to the Reppo network.Reppo’s Filecoin Models.
Reppo’s Filecoin, DIMO and Casper Studios’ partnerships were covered in the DePIN section.
Many thanks,