When NFTs are also programmable smart contract wallets, they become applications
ERC-6551 allows any existing NFT to have its own wallet, establishing an entirely new concept of “NFTs as Wallets.” That simple premise, combined with native support for Account Abstraction (ERC-4337), is transformative.
4337 (Account Abstraction) allows smart contracts to act as wallets and execute transactions on behalf of users (programmable). 6551 extends the use-cases of NFTs exponentially, allowing NFTs to evolve into a new type of tradable asset class, and setting the stage for a never before seen “Wallet Economy.”
Because 6551 is multi-layered (ie NFTs with wallets that can own NFTs with wallets), two dominant patterns emerge:
NFTs are now able to hold inventory (other ERC-721 NFTs and Tokens) in their own wallets
Wallets are now able to acquire abilities (NFT Applications with their own programmable smart contract wallets)
We will focus on the 2nd pattern in this litepaper, to outline our vision for NFT Applications (Smart NFTs), and why we think they are groundbreaking for the wider NFT ecosystem as a whole.
ERC-6551 (Token Bound Accounts or TBA) was proposed in Feb 2023 by Future Primitive. In a nutshell, ERC-6551 is an Ethereum standard that allows any ERC-721 token (NFT) to be bound to a smart contract wallet.

Let’s summarize some of the key features and benefits (paraphrased from this C3 Labs article):
ERC-6551 enables every ERC-721 NFT to own/hold an arbitrary amount of any other token (ERC-20, 721, 1155, etc.).
It doesn't require any changes to existing NFTs or any new NFT contracts to be deployed.
It "automagically" adds these accounts to all existing NFTs without the NFT owner needing to perform any transactions.
It is practically infinitely extensible - because the token bound accounts are smart contract wallets.
It natively supports Account Abstraction (ERC-4337)
It enables new "stateful" use cases for NFTs, including "sci-fi like" concepts such as autonomous on-chain Al agents
It's a major step in the evolution of NFTs taking them from simple PFPs to Tokenized Avatars.
While most coverage on ERC-6551 have focused on avatars being able to own static inventory, what’s most exciting is avatars being able to own, trade, and use dynamic abilities via NFT applications.
TL;DR - “NFTs we use vs. NFTs we own”
NFT Applications are ERC-721 NFTs bound to smart contract wallets with programmed logic (via ERC-6551). NFT apps can be used for anything a smart contract developer can imagine, with some use cases being Defi Trading Bots, Light Games, Quest Automations, DAO Rewards, etc.
NFT apps are 721 NFTs so they can also be traded or transferred just like any regular NFT.
We’re calling them Smart NFTs, or NFT Applications. NFT Apps can be used to power Networked Playable Characters, a play on NPC’s (Non Playable Characters) in games.
For example, in Bored Ape Yacht Club lore, there is a fishing lake. With 6551, a Bored Ape can deploy its own wallet (this is possible today). Imagine a fictional NFT application called “BAYC Fishing Pass.” Only Bored Ape holders who own a Fishing Pass NFT can “fish.”

Every day that they connect their token bound account to the BAYC website, a random amount of $FISH tokens will be airdropped into their Fishing Pass NFT’s sub-wallet (as programmed by its smart contract). There is a finite supply of “BAYC Fishing Pass” NFTs that exist, making these a valuable commodity within the BAYC ecosystem. BAYC holders can choose to sell their Fishing Pass NFT with its own TVL (Total Value Locked) i.e. everything in its wallet, or transfer its inventory out and just sell the Fishing Pass NFT as an empty wallet.
With this simple design, BAYC now has a way to incentivize hodlers to visit their website through gamification, whilst extending their lore, whilst rewarding engaged owners. And they can do this through a simple NFT application with a light UI, which takes much less effort to build than an entire Otherside metaverse.
But this barely scratches the surface of what’s possible.
What if we integrated the NFT application wallet with a general purpose AI that could execute Smart Contracts for or against users? Then we could create a more complex fishing game where the best strategies generate the most $FISH. The player sets their strategies, the smart contract executes, with an AI adapting to environmental feedback. Now we can start imagining static PFP’s becoming AI-automated Smart Agents that can play a game for or against users.
Smart Agents will bring forth the advent of an interoperable gaming universe, effectively improving existing platforms like Treasure. Through Smart Agents and their interoperable inventories (6551), meta games can enhance in-game economics. For instance, introducing "immigrant" characters from other universes to add new layers of tasks, trades, and in-game asset dynamics.

NFT applications represent an inflection point: “NFTs we use vs. NFTs we own.” Conceptually, NFT Applications smash the scope of current NFTs, largely limited to artwork and digital collectibles, and introduce myriad possibilities for new use cases within Gaming, Defi, Investing, Memberships, DAOs and more.
We won’t cover all of these use cases in this litepaper because others have already written about them. But we’ve linked to some of the best articles in the appendix.
TL;DR - “Unlearn to Learn”
ERC-6551 is gaining strong adoption amongst developers, and can be applied retroactively to all NFTs in existence. NFT marketplaces like Opensea are gradually rolling out support for 6551 NFTs. Thus far that has been limited to iFrame support, allowing NFT creators to add UI (like drawers) for NFT owners to see their NFT inventory.

Fundamentally, because NFT applications are a new archetype, a new form of marketplace experience is needed that is more akin to an App Store / Wallet marketplace than an existing NFT marketplace. In that sense, it will be more similar to Whop or even Shuffle than OpenSea. This can be a challenging concept to grasp because it requires a different mental model of NFTs for consumers, more focused on utility vs speculation.
Some of the key features an NFT app store should have:
State Handling: NFT applications are Stateful as opposed to Stateless, i.e. when a seller sells an NFT with TVL, the buyer should receive that NFT with the same TVL that was listed.
Account Management: NFT applications will need additional UI to manage Owner-Controls: Things like application settings, equipping, transferring assets within sub-wallets, stack management, etc. NFTs as we currently know them are Creator-Controlled, where an NFT’s features (e.g. traits) are set by the creator upon initial deployment.
In-app Revenue vs. Royalty: NFT applications will not follow the same business model as artwork and digital collectables where the intermediary’s revenue comes from a cut of the initial sale and secondary sale (modeled after the art world). NFT applications will be able to generate in-app revenue, and the majority of NFT applications will be distributed for free (App Store model).
In addition, an NFT app store should emphasize distribution and composability, over liquidity. It can do this by offering light contract templates and SDKs for developers (similar to ThirdWeb, Open Zeppelin, etc.) to leverage default 6551 implementations rather than starting from scratch. It should also provide a no-code launchpad where non-developers can also release their own NFT applications, and charge mint fees (similar to Zora). Working off official templates will allow NFT app developers to build apps that are more composable, secure, and affordable, and leverage plug-n-play mechanics.
Ultimately the NFT app store creates value (and generates revenue) when NFT apps themselves accumulate TVL and generate revenue.
NFT applications are composed of four layers: a presentation layer, logic layer, transaction layer and blockchain.
The Presentation layer presents the NFT Metadata in a format that is easy for users and other applications to understand.
The Logic Layer can be built into an on-chain smart contract, hosted as a rest API, or so complex that it forms its own trained AI model.
The Transaction Layer is linked to the Logic Layer and used by the NFT Application to execute actions on the Blockchain, such as buying or selling.
The Blockchain Layer is where transactions are executed and where the Presentation Layer receives metadata, while Logic analyzes state.

Here are 3 examples of NFT Applications, showcasing different use-cases:
Beep is a simple proof-of-concept we’ve built to showcase a DCA (Dollar Cost Averaging) trading bot as an NFT application (Defi usecase). Beep periodically exchanges one token for another on a set frequency. Trading bots in the form of NFT apps are safer than current Telegram trading bots, as keys are managed by your regular main wallet (and transactions are initiated using permissions built into the account implementation).
https://opensea.io/assets/0x9cAc72EFe455ADb4f413A8592eD98f962B7bE293/0
Presentation Layer
An iframe that dynamically updates as the balances of tokens after each transaction.

Logic Layer
The logic layer is user selected frequencies and amounts that are then regulated by AWS microservices. For Beep specifically, we’re using AWS but the logic layer can also be regulated onchain per the developer’s preference (see Food Scramble below)

Transaction Layer
The logic layer regulates the frequency and amount of each transaction and calls the transaction layer to execute the token trade. Currently, the Beep Bot uses a custom 6551 implementation. However, with Future Primitives’ launch of V3 account implementation, we believe that, in the future, the transaction layer will consist of an account implementation extension contract that interacts with the base 6551 implementation. This methodology will allow for more interoperability with dApps and crypto wallets.

A Networked Playable Character (NPC) is a specially trained AI personified as an NFT. The NPC can be trained to play within a game to interact with other characters, collect game items, or complete quests. AI NPC’s will be used in Parallel Colony
Presentation Layer
The presentation layer is a dynamic iframe that provides information about the activities of the NPC as well as any items collected.
Logic Layer
The logic layer is an AI model that has been trained by the owner of the NPC. The owner sets up initial parameters for the AI and then the AI determines which actions to take.
Transaction Layer
The AI-powered logic layer determines if items should be claimed, traded, or sold and then initiates those actions through a transaction layer.
Food Scramble by CodeChef is a NFT application where players have to collect tokens that represent hamburger ingredients in order to complete hamburger orders. Food Scramble is 100% on-chain, which means all the NFT application layers are governed by smart contracts. It’s a great example of some of the interesting experiments that are happening organically in the Tokenbound community. https://github.com/codechefsong/Food-Scramble
Presentation Layer
The presentation layer is a dynamic SVG image that updates based on ingredients in the TBA as well as the order completion status.
Logic Layer
The logic layer is a smart contract that controls when tokens can be claimed or airdropped as well as when players can obtain rewards based on the completion of burger orders.
Transaction Layer
The Logic layer determines a player’s rewards and then initiates an airdrop function through the transaction layer that deposits the rewards into the application’s TBAs. Likewise, when ingredients are spent or items are used up by players; the Logic layer will initiate withdrawals from those accounts.
Given the vast potential and diversity of NFT applications as showcased above, there's a growing need for standardized building blocks to expedite development and maintain uniformity. This is where Light Contract Templates come in.
These templates serve as foundational blueprints, allowing developers to swiftly implement, maintain, and deploy NFT applications while ensuring consistent interactions across different platforms and systems. Below is a high-level overview of what a light contract template might look like.

Utilizing such templates not only streamlines the development process but also augments the security, affordability, and composability of NFT applications. Leveraging default implementations and plug-and-play mechanics, developers can now focus more on innovation and less on reinventing the wheel.
ERC-6551 opens up a new world of possibilities but also requires additional standardization and systems to ensure interoperability between different NFT applications.
As a greater number of NFT Apps are built with more and more complex logic, we will see more on-chain assets connected to off-chain backends or logic layers. Similar to the various levels of centralization one would find for metadata storage, there will be various levels of centralization in the logic layer.
This is no different than today where a PFP creator can decide to generate 100% on-chain metadata using base64 encoding, upload their metadata to IPFS which is more centralized, or host the metadata on a website server 100% off-chain. The NFT logic layer can be a network of smart contracts that automate calculations or the contract can leverage oracles to automate 100% on-chain transactions. Here is an example provided by Chainlink.
We’re interested in the convergence of 3 key trends.
Growing developer adoption of ERC-6551
Sentiment shift towards Consumer Crypto Applications
Modularity as an emerging product model
ERC-6551 developer adoption
ERC-6551 was only launched in Feb 2023 and already there are over 1.6K developers in the Tokenbound Working Group. And as of September 21, 2023, a total of 13,152 TBAs have been generated across all EVMs. The community is active, engaged and growing fast. You can view real-time stats on this Dune Dashboard.

Popularity of Consumer Crypto Applications
In the past few months, 2 of the most successful projects in the crypto space have been Applications: Unibot and Friend.Tech. At a point in September, Friend.Tech’s trading volume was higher than the total trading NFT volume on Ethereum. And as of Sept 14, the app earned $15.5 million in fees and $7.7 million in revenue.

While Friend.Tech may not necessarily be sustainable, it validates a shift in crypto sentiment towards application value capture instead of protocol value capture. Similar to Angry Birds, Fruit Ninja or Vine, applications like Friend.Tech, Steal Cam, or Unibot, might only need to capture attention and revenue for a short period of time before fizzling out, to be replaced by the next popular application on the trending charts.
Product Modularity
Finally, an emerging product pattern worth noting is Modularity. Last week, Metamask released Snaps, which allows 3rd party developers to build external add-ons, extending Metamask’s feature set. We see NFT applications working the same way. If your metamask wallet owns an NFT application that supports ERC-4337, you no longer need to onboard to a different Account Abstraction wallet, your metamask wallet has now magically gained 4337 features. With NFT applications, any EOA wallet can become a super-app, not just Metamask.
In fact, modularity is a key pattern behind most super-apps - an app combining multiple services into one. China’s super-app WeChat has 1.32 billion monthly active users, equating to about 80 percent of China’s total population. Within WeChat, external parties can create Mini-Programs for everything from booking taxi’s (using Didi’s Mini-program), to shopping (using Uniqlo’s mini-program). In China, WeChat mini-programs are more commonly used than standalone websites or apps. Modularity which folds into multi-purpose functionality provides an ideal environment for network effects through enhanced UI and SSO (Single Sign On).
Similar to how NFTs (enabled by ERC-721) were a major catalyst of the last Crypto bull cycle, NFT applications (enabled by ERC-6551) could propel the wider market forward by enabling new use-cases that were not previously possible. Fun fact: Benny from the Future Primitive team was part of the original Crypto Kitties team that authored ERC-721.
ERC-6551 is the missing step to turn accounts from owned assets into traded assets, creating an entirely new on-chain wallet economy to facilitate the buying and selling of smart contract accounts (and wallet abilities)
These wallet abilities are what we are calling NFT applications (for now). We believe a new type of marketplace is needed to support NFT wallets and wallet abilities as a new tradable asset class.
This is ultimately what we are building with Kiosk (tbkiosk.xyz) - “The App Store for Smart NFTs.” Key challenges we face are: The ERC-6551 standard is still just a few months old, the current NFT market still predominantly revolves around speculation on artwork and digital collectibles, and non-Crypto-native audiences dismiss NFTs as “overpriced jpegs.”
This is why we’re starting by rolling out a series of Proof-of-Concept (PoC) NFT applications to demonstrate what’s possible, in collaboration with some of the leading projects & communities in the space (to be announced). In order to ensure the success of NFT applications, Kiosk will provide the building blocks for projects to launch and maintain NFT applications that can be viewed, interacted with, and traded in a decentralized manner.
We’re excited to work alongside the Tokenbound community and look forward to connecting with creators and developers already exploring (or are interested in exploring) ERC-6551. Let’s build the next era of NFTs together.
Get in touch:
Email: info@tbkiosk.xyz
Join the Tokenbound working group: https://t.me/tokenbound
Community Paper Version 0.1. Written In Public As A Living Document:
Comments, Contributions & Reviews from: Dee Wu, Bernard Y, Adrian Lai, Audrey Akwenye, Arjun Guru, Jim Sexton, Fancy, Serge-Raymond, Sam McCarthy, Paul Gadi, Reggie Ba-pe III, Ravi Agrawal, Blake Kim, Mind Apivessa, Benny Giang, Joel John
Appendix:
References & Resources
https://medium.com/@lootadventureweb3/erc6551-is-the-future-bbc2b69f8e2b
https://newsletter.auditless.com/p/al-013-the-wallet-economy-is-here
https://research.cr3labs.com/p/what-are-token-bound-accounts
https://watcher.guru/news/friend-tech-generates-more-fees-than-bitcoin-solana
https://eips.ethereum.org/EIPS/eip-6551https://tokenbound.org/

