***Overview***

𝔹asedAI is being developed by the PepeCoin team and Based Labs as an AI L1 blockchain featuring full ZK privacy and smart contract capabilities. Documentation exists but is incomplete (some details are being released over time with each testnet phase).

𝔹asedAI whitepaper: https://arxiv.org/abs/2403.01008
Documentation: https://docs.basedlabs.net/based-cli-documentation/cs8Tej4WRmZe0K7QqRFH
Additional information: https://research.pepecoin.io/

See the whitepaper, but in brief, there will be a maximum of 1,024 "Brains" whose operators point to a particular LLM, set proportional distributions for payments to attract miners and validators, and set payment requirements for end users to submit jobs to the Brain.

The monetary unit will be $BASED (currently $basedAI), which is inflationary on an annual halving schedule. Over time, the implicit idea is for network emissions to be increasingly exceeded by user payments in $BASED.

But for the outset, the PepeCoin team estimates 30,000 to 80,000 $BASED annually to each Brain operator. There is some confusion about whether this is the total amount to the node (i.e. in which case approx 75% of this amount would be distributed to miners and validators, with 25% going to the brain operator) ot to only the Brain operator. But either way this is a lucrative operation given the current price of $basedAI and especially with the possibility of $BASED increasing in value in the future if the 𝔹asedAI network is successful.

The intent is for Brain owners to not need deep programming skills to operate their Brain effectively. But, even the lighter skill required is intimidating for many, and effective management to maximize yield (e.g. by pointing to LLMs that will generate the most business and adjusting parameters) may also be too active for some brain owners even if they have the technical operation capabilities.

This presents a recurrent and potentially highly lucrative business case for Empyreal to manage a Brain Operation Service in exchange for 10% of earned $BASED.

Furthermore, this operation could in turn be leveraged for Empyreal B2B clients as an SDK service, i.e. with the ability (notes below on how to price this) to have a Brain pointed to a defined LLM for the benefit of their protocol functions, fitting with Empyreal's AI-related SDK modules and service offerings.

***Development needed:***

Smart contract A: For Brain NFTs. One way for users to acquire brains is to burn PepeCoin for brain credits and use these credits to mint a Brain NFT. The NFTs will be composable. These brains are immediately activated. Contract needs to:
- accept Brains as deposits
- have functionality for Empyreal to be able to set 𝔹asedAI parameters for the Brains but without ability to move the brain out of the smart contract
- Owner ability to claim accrued $BASED and withdraw NFT
- Probably want to stipulate that onwers may not get the same brain NFT when they withdraw (this would preserve the ability to have exits done in some preferential order, e.g. preserving high-functioning brains for the overall pool)

Smart contract B: 100k PepeCoin stakes. The second way to acquire brains is to Stake 100k PepeCoin for 90 days, after which brain functionality is activated. On unstaking, the brain turns off, but no PepeCoin are burned. This would need to have same functionality as above but tracking time and no pool rewards for first 90 days.

Smart contract C [optional]: Pool of PepeCoin in less than 100k chunks. For each 100k accumulated, a staking transaction takes place, kicking off the 90 day wait. Rewards would be anticipated to be lower due to the inefficiency involved, but many people ask for this service. Smart contract for this is a bit more complex.

***Operation***

The idea would be to optimally balance (and regularly re-balance) the LLMs pointed to by the set of Empyreal Brain Service Brains, and set other parameters, to maximize returns for pool participants. With active management, the service can likely make up at least part of the fee paid by users vs. what how well they would likely have done on their own.

Fee: 10% of earned $BASED is revenue to Empyreal.

Suggestion to Empyreal: Consider holding 10% of the earned fees as $BASED for staking/ validating compounding revenue (in addition to the additional revenue, this Empyreal-controlled tranche could be used advantageously to help maximize revenue for the Empyreal Brain Service Brains)

***B2B clients***

Great opportunity for Empyreal to market the ability to point a Brain or multiple Brains to customer- or partner- identified LLMs to benefit their business. In addition to whatever service charge is billed to the client by Empyreal (or included as part of a broader package), the effects of these arrangements on Empyreal Brain Service customers need to be considered. Thus:

- If the custom LLM configuration results in revenue that is under average earnings for a Brain, the customer pays the difference in $BASED and this is added to the earnings pool for Brain Service participants
- On the flip side, if the custom LLM configuration results in *more* revenue than the average Brain, then that customer would receive 50% of the over-average amount in $BASED.

***Outlook***

Folks are _constantly_ telling me that they want to operate a brain but are worried about their technical ability to do so. I don't know how many of these users would actually follow through to the point of taking advantage of an Empyreal Brain Service, but I would conservatively estimate the Service attracting 50 Brains. I can personally guarantee at least 10 from my own NFT/ PepeCoin staking deposits, and I would of course help advertise the service.

I do think that striking quickly is important here. Rewards will be proportionally larger at first before more brains come online, and the market share is likely to be reasonably sticky, especially if Empyreal can demonstrate that it is able to outpace what most users could achieve with less active indiviudal management of their Brains.