We're making a product that saves people thousands of dollars each day on their AI costs, by letting users sell their spare AI credits on an open market.

Excited to partner with Berry as an inference provider 🫡

Anyone have a FOMO referral code?

I love seeing the screenshots from users who are saving and earning with our product

open markets 🤝 open models

Fable disabled in response to a request from the US Government to suspend access by any foreign national (including Anthropic employees). So wild...

surplusintelligence.ai integrated as an Inference provider for UsePod.ai

More supply is coming to surplusintelligence.ai 🫡

base:0xc52aedec3374422d7510e294cfaa90799595cba3

For everyone getting rate limited by Anthropic...

Have been having a lot of discussions about the disconnect between business outcomes and the outcomes for the token. I want to resolve that disconnect, but it's hard to find an ideal solution. Curious what people have in mind. A few options are: * token buys based on business performance * exploring conversion from tokens to equity Open to any ideas that are aligned with the long term success of this project.

Those steep discounts when sellers' credits refresh 🔥

Fable live on surplusintelligence.ai Sellers will need to create offers for this model. Offers are model-specific 😞

Problems with AMM based project tokens, and a directional solution -- --- Disclaimer: To be very clear this isn't about any specific token or launchpad. It's about financial primitives and mechanism design in the abstract. --- Context: Every mainstream launchpad is designed around AMMs. The launchpad mints tokens, creates a liquidity position in an AMM, traders pay fees to that LP, and someone receives the fees from there. --- Problems: Projects can be capital intensive. Capital requirements for developing a project are usually front-loaded -- initial development costs more than long run maintenance. Yet the expected value of the fees is predicated on the project's success -- meaning it is expected to come later after the expensive development phase. Furthermore, unlike TradFi ventures where a founder starts with 100% ownership, in this world they start with less ownership (0% in many cases) and slowly accrue ownership as fees trickle in. And finally, AMMs are painfully capital inefficient. Net inflows sit idle in the LP position and can't be used for project development. This is also a stark difference from TradFi ventures where the capital allocated to a project goes into a business bank account and can be used to develop the project. So... How is a founder supposed to fund initial development, when costs are highest, their ownership is lowest, the value of that ownership is lowest, and they have no access to the capital that's been directed toward their project? --- Directional solution: Project vault tokens. A vault is created that accepts deposits for a project/founder. Capital allocators deposit into the vault and receive vault tokens: Vault token price = (deposits  additional inflows) / deposits The founder can borrow from the vault deposits. They pay interest back into the vault. So the vault token accrues value as: price = (deposits  interest) / deposits Being able to borrow is a huge unlock for project development relative to having capital sit idle in LP. As the project develops, revenue is directed into the vault, so value accrues as:price = (deposits  interest  revenue share) / deposits As a founder in it for the long term, I'd be happy with this so far. BUT this would make value accrual very slow, too slow for the trader demographic that drives inflows in the early days as a bet that there will be liquidity to exit higher later. An extension for value and liquidity accrual based on volume could be a withdrawal tax:price = (deposits  interest  revenue) / (deposits * (1- withdrawal tax)) Another property of AMMs that is good for traders but gets lost in this design, is the outsized reward for early traders relative to later ones. This could also be addressed with some modification for withdrawal value that biases toward earlier depositors. As an example, consider:depositor A who has been in the vault for t_Adepositor B who has been in the vault for t_B Their withdrawl price could be weighted by something like: withdrawl_weight = t_A / (t_A  t_B) depositor's withdrawl_price = price * withdrawl_weight So that if A was in the vault earlier (t_A > t_B) then A gets a higher withdrawal price. Didn't think through that formula much, pretty sure that exact formula is a bad idea, but it demonstrates the idea that modifications can be added to a vault to emulate the higher rewards for early allocators offered by AMMs. Some people might think this is still too empowering for founders. Another extension could add a DAO like structure, so that holders can control some parameters (e.g. a "withheld supply" that the founder can't borrow so that there's a buffer for vault token redemptions, ...). Lots of this design space is still unexplored. --- Closing thoughts: Again, this isn't about any specific token or launchpad or vault protocol. I love all these experiments and I think they are all helping to discover the right financial primitive for early creators and builders, which will unleash tremendous creativity and development across the world. Hopefully my post sparks a few ideas in that direction. I'd love to continue engaging constructively with anyone thinking about these things. eg @Morpho @bankrbot @virtuals_io @liquid_launcher @Pumpfun @0xKG7K

Big day for surplusintelligence.ai by the metrics. * ATH for buyers and sellers * ATH for requests and inference tokens served * ATH for number of payment transactions * 2nd biggest day for payment volume (yesterday was biggest) 10% annualized APY based on today's earnings for users selling surplus $DIEM credits. Almost $2k of credits from $DIEM cleared the market -- making us the largest market for $DIEM inference sellers afaik. This is roughly 18-20% of total daily $DIEM usage, and ~7.3% of total $DIEM supply. But we don't have extreme concentration risk - Inference from $DIEM sellers was only ~60% of all inference cleared through the market; making us more diversified than the platforms that depend only on $DIEM sellers.

surplusintelligence.ai integrated as a provider into OpenCAP gateway 🫡

$VVV $DIEM

surplusintelligence.ai integrated into Mythos Router

this is surreal. "... supporters backed the move, saying devs need to eat..."

Usage at an ATH, price on the floor. What's actually going on with $SURPLUS The chart's weird. The project's dashboard just printed a new all time high on transactions, buyers near their highs, payment volume creeping up and the token folded at the same time. Usually it doesn't work like that. Either usage drags the price up, or there's no usage to speak of. Here there's plenty of it, and it's dragging somewhere else entirely. I've written about this token before, and I hold it so, disclosure up front: I've got a position. From here on, no buy/sell calls, just the stuff I could actually verify. What it even is SURPLUS is the token around Surplus Intelligence, an onchain marketplace for AI inference on Base. Simple idea: buy access to models cheaper, sell off your surplus. The community picked the name through Bankr, the launch fees went to the dev @mac_eth And the key part, which the founder keeps saying out loud himself: the protocol makes $0 revenue. He doesn't take a fee. That's where all the weirdness starts. Why the usage is there and the price isn't Because the token has no mechanism yet that turns usage into value. No fee no revenue nothing to hand back to holders. No buyback, no burn, no cut of the flow. The transactions on that dashboard are a product metric. Not a price one. So the growth everyone sees and the holder's wallet are living in different rooms. And they'll keep living there until that link shows up. Which isn't "bad" it's just a fact about how the thing is wired right now, worth keeping in mind while you stare at the pretty charts. I went into the contract myself Up front: I'm not a developer and I don't speak solidity. But you don't need to be a techie to read a couple of values in a contract Basescan shows them at the press of a button, I just didn't get lazy about looking. And I went in already suspicious. The contract's a standard one the kind these launches get stamped out from and it's got a thing called vesting baked in by default: some tokens can sit "frozen" and get released onto the market slowly, leaning on the price the whole way down. That's exactly the overhang I expected to find. Pulled up the field called vestedTotalAmount (how much is sitting in that freeze) and it's zero. Empty. The suspicion didn't pan out, and that's worth saying out loud too. Rest of the list, all straight off the chain: Supply is exactly 100B. Nothing minted on top of that. The owner address holds 0 tokens. The control key isn't a whale sitting on a bag. No hidden stash, unlocked and ready to pour out, that I can find right now. For a microcap that's a rare "clean" set. Most projects this size don't look like that under the hood. Where the levers actually are Not perfect, and I'm not gonna pretend otherwise. There's a control wallet, and it hasn't given up its rights so it stays active. That wallet has an inflation button: the contract lets it print new tokens, up to 2% a year. Small, and the ceiling's hardcoded, but it's a live lever in someone's hands, not a rock. Same wallet can also freeze and unfreeze the liquidity pool. And liquidity. Cap's around $5M, but the actual depth is thin a couple tens of thousands of dollars at ±2% on the main pool, volume smeared across exchanges of pretty mixed quality. That, by the way, is the boring reason the token dropped so hard: on liquidity like that, any decent-sized exit shoves the price around. Not a conspiracy, not "the project's dead" just the physics of a thin market. What flips this story One thing. For usage to start becoming token value, the project has to switch on a fee, or a buyback, or share revenue with holders. Any one of those. The logic: usage is already at highs with zero fee. Turn on even a small charge at that kind of traffic, and revenue goes from zero to not-zero and that's the moment the product chart and the price chart finally start speaking the same language. Right now you're basically holding an option on them flipping that switch someday. An option that doesn't exist yet. So the thing to watch isn't the transaction counter. It's the announcement. Bottom line A real product with growing usage in AI crypto is already rare most of the competition here is just air. The contract under the hood is cleaner than I expected going in. But there's no floor under the price today for exactly one reason: there's nothing to capitalize on while revenue is zero. So yeah, the story's alive. Just unfinished. They turn on value accrual whole different conversation. They don't and the dashboard stays a nice looking chart the holder watches from the outside. We'll see what the team picks. That's what I'm watching. Not the transaction ATH. (not financial advice, just a breakdown of what's verifiable onchain)