Roadmap
Roadmap in no particular order or timeline
This roadmap may change as development continues
Decentralize GPUs
Launch a peer-to-peer network to support a distribution of interconnected GPUs that host machine learning models, such as LLM's and stable-diffusion.
Complete
Add first model
Integrate the open sourced LLAMA2 model
Complete
Launch
March 11, 2026 - Launch of a minute amount of tokens
The launch will take place on Ethereum for a percentage of between 84-90% of est. total supply.
Complete
Deploy blockchain Devnet Vitalik
Deploy the first version of devnet.
Encorporating:
→ Peer (model hosters) requirements
→ Simple rewards mechanism
→ Peer consensus mechanism
→ Computation based rewards mechanism
Complete
Deploy blockchain devnet 2.0
Deploy the second version of devnet.
Encorporating:
→ Peer (model hosters) requirements
→ Simple rewards mechanism
→ Peer consensus mechanism
→ Computation based rewards mechanism
→ Democracy pallet encorporating for model vote-ins and admin functions
Est:
Q3-Q4 2024
Deploy blockchain devnet 3.0
Deploy the third version of devnet.
Encorporating:
→ Peer (model hosters) requirements
→ Simple rewards mechanism
→ Peer consensus mechanism
→ Computation based rewards mechanism
→ Staking & Validating
→ Democracy pallet encorporating for model vote-ins and admin functions
→ Treasury pallet encorporating for managing slashed funds
Est:
Q3-Q4 2024
Deploy blockchain testnet 1.0
Deploy the first version of testnet. This will be the testing grounds for mainnet.
Est:
Q3-Q4 2024
Testnet Contributions and Upgrades
Open Testnet for contribution from the community, test the network, and continuously upgrade the network.
Est:
Q3-Q4 2024
Inference Consensus / Inference Validation
Develop an optimistic approach to inference validation.
Est:
Q3-Q4 2024 +
Upgrade P2P Network
Make any neccessary upgrades to the peer-to-peer model hosting network
Est:
Q2-Q4 2024 +
Deploy blockchain mainnet 1.0
Deploy our mainnet blockchain.
Est:
Q3 2024 - Q1 2026
Developer Products
Launch a suite of developer products, such as an API, to give community members the ability to build applications on top of the network. As well as giving assistance in accomplishing big ideas.
Est:
Q3 2024 - Q1 2026
Consumer Products
Alongside innovating and building out the blockchain (seen in next card), we will heavily focus on developing consumer products around the current AI models alongside community developers, such as building easy to use chatbots and training interfaces. This will be anything from agents to help businesses operate to mobile applications to designer tools for applications like Photoshop and GIMP.
Est:
Q4 2024 - Q1 2026
Federate Learning
Implement federated learning to the subnets.
Est:
Q1 - Q4 2026
AI Interconnection DAO
Build a DAO for the sole purpose of bringing interconnected AI to the masses as well as educating those interested in the space.
We hope this contributes to the future of AI being in the hands of the people as opposed to the few,
Est:
2026
Bring Everything Onchain
Bring as much of the machine learning network onchain, such as the training, data, and other facets.
Est:
2026
Layer Boxes - Mainnet 2.0 - HyperMolt
Create Layer Boxes directly on the blockchain. This will be a pallet directly on the blockchain with the ability to mesh multiple models into one "Layer Box" for the best optimal outputs.
The thoughts behind Layer Boxes is similar to that of an aggregator. It will allow end-users to consume AI models and receive the absolute best output through optimized routing algorithms.
Est:
2026
Adapt, scale, and build
Listen to the community wants and needs, adapt to current and future market conditions, and continously build to ensure peer-to-peer AI supercedes centralized AI
Est:
Forever