Branche Whitepaper

A New Paradigm of Semantic Privacy-first DID Protocol

Branche aims to create a semantic privacy DID protocol to achieve higher value user privacy protection and data transactions. This whitepaper will provide detailed information on Branche's technical architecture, user profile generation, data authorization, and benefit acquisition.

Branche's core technologies include semantic privacy decentralized identification (DID) protocol and AI browser plugins. Through the DID protocol, Branche will achieve decentralized data management, providing a higher degree of privacy protection. The AI browser plugin can help users automatically capture their own interaction behavior and knowledge graphs, further forming their own user profile.

In Branche, users can authorize their data through zero-knowledge proof (ZK) to obtain certain benefits. This data authorization method not only ensures user data privacy but also allows users to benefit from it.

    In summary, Branche is dedicated to providing users with a highly private, decentralized, and potentially valuable data engine.


    Branche's Vision and Goals

    To help people retake personal data soveignity and empower users with personal AI with personal data.

    Branche aims to create a semantic privacy DID protocol, hoping to help users regain autonomy and choice over their data. In the world that Branche hopes to create, everyone has true control over their data and can choose whether to authorize their data or revoke data permissions at any time, while also obtaining benefits through authorized data access.

    At the individual level, Branche aims to help users turn their data into a user-defined, personal AI assistant (the second brain or VA), which can assist users in message recommendation, filtering, and processing. These are services that were previously controlled by large corporations and were unable to be personalized, but often have the most significant impact on individual autonomy.

      Additionally, through our underlying data protocol, we aim to introduce zero-knowledge proof applications into the web2 world, creating a seamless web3, semantic web of data-as-identity. By using zero-knowledge proofs, we aim to create a new social contract, reducing trust costs, verification costs, and performance costs, ultimately providing identity-programmability for network states.


      Core Advantages and Features of Branche

      The core advantages of the Branche protocol are three-fold: "semantic, zero-knowledge, and permissionless". We are building an internet protocol with the following specific meanings and characteristics:


      • Branche semantically represents the data of each user, which means that a specific user's ontology description and behavior logic can be expressed in natural language and semantically represented in vector data. In the future, data can be retrieved through natural language description, resulting in a more readable, scalable, and usable data that is like a language, independent of all ecosystems (natively cross-chain), and exists in all ecosystems at the same time.


      • The composition of all user identities is achieved through a set of de-identification, non-directional, and privacy-protective data structures. Users can call their data at any time for self-identification, social trust building, knowledge proof, etc., to prove their own data. Through embedded zero-knowledge, data sovereignty is ensured, and in the future, users can even prove their data through zero-knowledge language models, creating "automated trust".


        • Under the Branche protocol, all interactions are native and permissionless. Data itself is independent of any existence and exists in any storage space in the form of a DID file. Users can store it locally, use proxy storage, or centralized custody. Users can retrieve their data proof anywhere, anytime, without any permission from any centralized service provider, proving and verifying their own identity.


        To provide so, comes a personal AI for everyone: Branche AI

        The core advantages of Branche AI lie in three main features: "personalization, autonomy, and versatility". We are building an AI that is unique to each user, with the following specific meanings and features:


        • Branche visualizes each user's AI as an extension of their own consciousness. Users can bring their intentions and knowledge background to AI through usage. In daily use, AI can provide personalized services such as custom recommendation algorithms, filters, etc. through its second brain memory, helping users regain their right to choose from enterprises.


        • Branche AI is independent of any large language model services. Users can activate Branche AI using any language model (even future local models) of their choice without permission. Users themselves control all components of Branche AI and can operate Branche AI as their own assistant, becoming their own agent.


          • Branche AI adopts semantic behavior data and knowledge graph parsing algorithms, applying LLM's generalization ability to users' daily data flow, achieving cross-platform generalization ability. In the future, it can even solve the problem of limited mobile phones and ineffective proof generation in real-life scenarios by accessing intelligent devices (OCR/multimodal).


          How does Branche achieve a decentralized data engine?

          The Branche protocol's data engine is designed to be decentralized with the use of zero-knowledge (ZK) technology. This technology allows for each piece of data to be digitized and have proof of data generation uploaded to Mina through Snarky. This approach offers several advantages, including:

          1. Lightweight: The use of ZK technology allows for data to be compressed and stored in a more efficient manner, making it easier to manage and transfer.
          2. Privacy application: The original data is abstracted and de-identified through ZK, which ensures that the owners of all data cannot be queried. This protects user privacy while still allowing for data to be shared and utilized.
          3. Off-chain smart contracts: By using off-chain smart contracts, Branche is able to reduce the cost and processing time associated with on-chain smart contracts. This approach allows for faster and more efficient data processing while still maintaining the security and integrity of the data.

          Let's explore an example to understand the concept better. Consider an advertiser who is interested in finding user data of females who use web3 in China. With the use of zkPoH, the advertiser can be assured of the historical proof of the data and the existence of its owners. The subsequent interaction can also confirm that the interaction object is the rightful owner of the data through proof. However, the advertiser cannot know who the owner is due to the privacy-preserving features of zkPoD. This feature ensures that the advertiser does not need to pay before reaching the other party, enabling a new transparent yet private data engine. These privacy-enhancing features are critical in many areas, including finance, healthcare, and personal data management.

            In summary, the Branche protocol's use of zero-knowledge proof (ZKP) technology and off-chain smart contracts helps to achieve a decentralized data engine that is lightweight, private, and efficient in processing data. This is because the ZKP technology ensures that data can be verified without revealing any information about it, thus enhancing the privacy protection. In addition, the off-chain smart contracts allow for faster and cheaper data processing, which makes the protocol more efficient. Overall, the combination of these technologies makes the Branche protocol a promising solution for decentralized data storage and management, with potential applications in various industries such as finance, healthcare, and supply chain management.


            Key Technical Components

            The core architecture of Branche consists of "Apps, DataPod, PKM, DMP, ZK-Provider, AI-Provider, Extensions", each responsible for the following functions:


            • This includes native applications developed by Branche, such as our first product: AI browser plug-in, which helps users automatically capture their interaction behavior and knowledge graphs, forming their own user profile. It also includes applications developed by the developer community or third-party developers.
            • Branche's future applications (names are tentative):
              • zkHandShake: An application based on zk that compares the data sets of two people to find overlaps. A typical example is that scanning a social code can obtain common friends/interests/behaviors of both parties.
              • zkDSP: A data buyer platform based on zk that helps with advertising placement and matching, finding users precisely, and proving that users have viewed ads without revealing their privacy.
              • zkJob: A job search/recruitment matching application based on zk that automatically matches suitable jobs/job seekers. And by verifying the knowledge background CAPTCHA using zk, it eliminates resume fraud.
            • Branche encourages developers to develop any applications based on the Branche protocol. By expanding privacy data to more application scenarios, Branche will provide more developer support and rewards in the future.


            • Branche's native data storage container is used for user data management, permission management, authorization signature, etc. DataPod itself also serves as a tool for user identity verification, providing a series of interactions with applications and data calls. Its significance is similar to that of AA-SCW (for data), helping users truly manage all their data autonomously.


            • The main function of PKM (Personal Knowledge Management) is to extract and index user data memory, reflected in the user's personal knowledge base, second brain, and knowledge management. In the future, users can authorize their data as an open knowledge base/robot for third parties, providing other users with contextual knowledge and generating income from authorizing their knowledge.


            • The Branche DMP data engine itself is a chain storage protocol consisting of the following elements:
              • Data processing module: All data is imported through AI-Provider on the user side and stored in the vector database through a data pipeline compatible with w3c and interact with dataPod for personal data CRUD.
              • Distributed vector database: The distributed vector database consists of different nodes (thema). Thema refers to different data topic libraries, which can be understood as each node having a theme. For example, the programmer node collects data related to programmers, and the Hangzhou node collects data related to Hangzhou. Such a database can be operated and deployed by curators, who collect and construct the recording library of different value groups. Of course, requesters can query across multiple nodes.
              • Data service API: Developers can query the public data of everyone on any node, and develop/analyze based on data, and access services to develop applications based on the Branche ecology.


            • Thanks to the support of Mina ZkApps, Branche's ecological applications can run in any chain environment and ensure privacy and data authenticity. Zk-Provider, as an independent existence, supports off-chain smart contracts and data interaction within the Branche ecosystem. We will provide more explanations for this part in the future.


            • When users use Branche's ecological applications, they need to access the LLM service to use semantic functions effectively. Users can choose to access the model through "local, official, or third-party" methods. Currently, we are researching technologies and papers related to large-scale language model encryption operations (embeddings cryptography) to ensure the privacy data of users is kept secure and will not be leaked when using large models. If you have any research in this area, please contact us at contact@branche.ai.

            Extensions (Ecosystem)

            • In the Branche ecosystem, all components are modular, and users can build Extensions based on the ecosystem to build more data ecological extensions on the basis of the native ecosystem. Branche encourages developers to develop any extensions based on the Branche protocol, providing more underlying extensions for privacy data. Branche will provide more developer support and rewards in the future.


              Ensuring user privacy and data security

              • Branche is an open data protocol that guarantees user privacy and data security through zero-knowledge proof. The data within the protocol has been verified by industry experts and professional code audit organizations to ensure code quality at the protocol level and protect user data security. However, as an unauthorized protocol, improper user operation may inevitably cause data leaks. We ensure that even if data leaks occur, the data itself cannot point to any specific user by abstracting users into each incarnation.

              • Branche's developer ecosystem also plans to increase the Bug Bounty program, working with developers to create a more secure and private data protocol. Branche will provide corresponding rewards for developers who provide bug reports and security optimization, committed to protecting user data. If you have any suggestions regarding Branche's security, please contact us at security@branche.ai.
              • We have always attached great importance to cutting-edge cryptography technology and its related technology implementation. If you are interested in joining Branche and participating in the development of cutting-edge applications such as ZK, TEE, and vector encryption, please contact us at hr@branche.ai.


                Capturing interactive behavior and knowledge graphs

                Branche's AI browser extension is Branche's first PoC product, aimed at demonstrating the potential of Branche. The Branche AI extension helps users crawl key information in the browser (Html, DOMevents, Content) in real-time for parsing and extracts key interactive behavior data and knowledge graphs through LLM. We will standardize a series of commonly used internet platform parsing and provide more website parsing for developer ecosystem users in the future as curators to capture more key information and help us set more tracking points and participate in data revenue sharing.


                Value of self-soveign identity portraits

                Personal level

                • All large free Web 2.0 internet products are essentially designed to let users use them for free, gain their data value, and sell the user portraits to B-side for profit. Thus, users cannot obtain ownership of their own data. From the user's perspective, obtaining their own identity portrait can establish their own credit and digital identity, serve as their own data authentication, and authorize their identity portraits (data) to obtain more services and benefits, and take control of their data without third-party supervision.

                Commercial level

                • In the past, advertisers or producers often had to pay for exposure to large traffic platforms because they did not have means to grasp traffic (interact with consumers directly). The more precise the user positioning, the higher the cost. However, if the traffic platform has a say in the data, advertisers often cannot determine whether their placement is reasonable and often only pay "protection fees" without addressing the root problem. Self-owned identity portraits can help businesses not need to pay for "protection fees" after going platform-free, and instead trade autonomously with users to increase certainty and value circulation.

                    Social level

                    • Branche believes that self-owned identity portraits and privacy can encourage more people to share their data, and more data will bring more confirmation to the community and create more value for society. When more people have self-owned identity portraits, it will greatly improve communication efficiency, reduce trust costs, verification costs, and performance costs, and serve as the underlying data of the new social contract paradigm. Its value lies in automating the process of establishing trust through active behavior, enabling web3 to move towards a future of trustful→trustless→trustless trust. In the future, we will conduct more exploration in the direction of Programmable Trust to increase collaborative efficiency.

                    Self-owned identity portraits help users create their own digital identities, which exist independently on any platform and are not controlled by any platform data. In the increasingly semantically ecological self-owned identity portrait, it will support more verification scenarios that were previously "offline/inefficient".


                    Zero Knowledge Proof (ZK) to authorize user data and achieve data security sharing

                    Using ZK technology can achieve secure data sharing by granting access to specific data without sharing the entire dataset. Users can prove their ownership of specific identities or attributes through generated VCs, allowing them to selectively disclose specific shared data without sharing their entire identity. ZK technology allows users to prove that they have the right to access certain data without revealing the content of that data to third parties. By using these technologies together, users can be authorized to access specific data without revealing the entire dataset.


                    Users rights & benefits of their data through authorization

                    • uthorization is described as "Who (Who) can perform What operation (What) on what resource (How)." We use the RBAC model to support users in creating different "roles," describing the semantic ownership of that role's data and setting rules for who that role's data is authorized to. Developers can also request data through smart contracts in the service and provide rewards/permissions to users after authorization (more: zkOauth, zkPoD).
                    • Rights: In the early stages, Branche rewards high-quality ecological behavior with tokens. After the user signs the Branche Everlast contract, 10% of the future value of all data will be used for the Branche ecological reward pool.


                    Branche's governance model and community participation

                      • Branche's governance system is mainly composed of two parts: Branche Labs and Branche DAO. Branche Labs is the core team composed of Branche's founder, core developers, and supporters, responsible for all early decision-making, development, and operation of Branche. Branche DAO is the organization that Branche hopes to grasp the core governance of Branche ecology in the process of ecological construction in the future, and the future protocol pricing, treasury, subsidies, incentives, etc. will be decided and voted on by Branche DAO to ensure Branche's Everlast. Branche DAO will set different thresholds for different users in the protocol and select agents through screening and voting. Any user in the Branche DAO community can initiate a proposal.
                      • Branche's core exchange contract will be deployed by Branche Labs' official multi-signature address and designed with timelock. Branche Labs will invite core contributors in the community to become multi-signature members.
                      • About Branche Everlast: Branche Everlast is a governance contract initiated by Branche Labs, inviting users to share their data according to the Exponential Decay. Part of the future data-generated revenue will be used to support the sustainability and liquidity incentives of Branche ecology. The core content includes decay rate, dynamic drip, and detention pool.

                      Decay rate

                      • Users can choose different Decay Rates, which can be understood as half-life, to share data revenue rights with Branche within the specified period.

                      Dynamic drip

                      • Due to the dynamics of the market and the regulation of market heat, the Dynamic drip will automatically adjust and reduce the excessive fluctuation of data prices to ensure market stability.

                      Detention pool

                      • The Detention pool is a livestock flood pool specially designed for Branche's sustainable incentives. By attracting the community to share revenue (voluntary taxation), it serves as an incentive budget for future Branche. The Dentention also serves as an automatic adjustment function for the market.

                        Currently, Branche is in its early stage and the technology selection as well as the ecosystem has not been determined. We are currently open to any ecosystem and Branche will greatly improve its scalability, making it applicable to more ecosystems in the future. Branche has already collaborated with dozens of Web3 projects to jointly build the Branche ecosystem and data privacy engine in the future.

                        Branche is committed to creating an open, private, and imperceptible Web3 data engine, reducing the cost of entry for Web2 users. If you are interested in partnering with us, please contact contact@branche.ai.


                        Branche's Roadmap and Development plan


                        • Branche's core roadmap consists of four major areas: "DevEco, Community, Growth, Buidl". DevEco will promote ecological development through community open-source projects and bounties, while the community will attract more people to join the Branche protocol and provide contributions. The growth part will serve the promotion and operational development of all Branche businesses, and Buidl's focus is on building the core technology of the Branche Protocol, technical development, distributed architecture, and protocol standards.


                        • Branche plans to create a multi-threaded development ecosystem, build a common symbol of value philosophy, and attract more people to participate in construction. If you are interested in joining us, please follow our latest news. Branche hopes to create a new paradigm for the common network, rather than just another useless web3 project. "Branchism" is a name that reflects our ideology of branching, and we hope to exist in a diversified and personalized form. If you want to learn more about Branche's philosophy, please read our blog.

                        Building and Deploying dApps with Branche

                        • Branche is not a Layer 1 public chain, but a data protocol, so any developer can use Branche's protocol technology and development tools to develop and release their own dApps. Branche's applications have features such as off-chain contracts, zero-knowledge verification, and self-owned data services, opening up new possibilities for seamless web3 applications. Developers can build their own scenario applications and services based on the Branche protocol, and jointly build a more complete privacy data ecosystem. If you want to learn more about Branche's development tools, please visit the Branche docs.


                          Challenges and Potential risks of Branche

                          • Branche is building a non-monopolistic data ecosystem, which naturally faces many difficult challenges and risks. The biggest challenge we believe is "technology" and "compromise": Branche faces many technical difficulties, including ZKP, TEE, MPC, P2P, DID, DMP, LLM, ZKML, etc., each with different difficulties and technical details to be overcome. We believe that we should persist in doing difficult but correct things;
                          • However, we believe that compromise is a more challenging existence. It is well known that the essence of Web2 Internet is to profit from selling user data. What Branche wants to do is nothing more than to break all the patterns of the "old world". Naturally, the road is very difficult and long. We can see that it is full of thorns. Whether it is facing the legalization of privacy processing or the siege of Web2 data giants, these challenges will inevitably be encountered in our visible future. Naturally, Branche's road must also make compromises for these difficulties.

                            We believe that the world is getting better little by little. Even if we just pull the world back a little bit from the disgusting end, it is something we hope to achieve. Even if the things we make are not as perfect as we thought, it is called "politics is the art of compromise". We are not sure whether we can call Branche a new form of revolution or just an idealistic joke. This will be decided by the people (or AI) who write history in the future; Branche's potential risks can be encountered in many ways, corresponding to the beautiful vision it portrays.


                            How does Branche compare to existing Web3 projects and solutions?

                            • As there are currently no projects that directly compete with Branche, we will compare it to the DID and AI tracks, although Branche is not limited to AI DID projects.
                            • The core of the DID track currently consists of three types of projects: 1. Open-source idealists 2. Web3 application developers 3. Web3 technology developers.

                            Open-source idealists:

                            1. The W3C DID project led by Solid (Inrupt) has limited use cases, outdated technology, lacks innovation and practicality, and can only serve the old ecosystem.

                            Web3 pragmatism:

                            1. There are currently many DID projects in Web3, but most of them focus on domain names, event platforms, identity aggregation, and PFP, and are expected to operate through airdrops without truly achieving PMF. Few of them use identity data technology to enter application scenarios.

                            Web3 technos:

                            1. Currently, we believe that zCloak is doing particularly well in terms of technology, and its zkID and validID are very good examples of zero-knowledge certificate technology implementation.

                            The AI track is mainly focused on LLM and GPT-based applications, where the former focuses on model-level innovation/reconstruction and the latter on constructing usage scenarios using existing interfaces/open-source models.

                              • We believe that we are working on a more fundamental protocol for LLM. In the foreseeable future, LLM will surely be a combination of local and cloud services, and will require more user context and taste adjustment locally. If LLM is likened to the language and logical ability of the brain, we are working on the memory and second brain, i.e., the external brain automation component.
                              • As for comparing with GPT-based applications, even the applications with many users currently have no barriers; most of them are just shells and some very traffic-oriented services. The cost of satisfying these needs is very low, and there is no competition threshold. We see that most of the users are just trying out or using the service for free.
                              • In the AI aspect, Branche emphasizes personalized AI and data autonomy. We aim to transform from the profit-oriented enterprise intelligence of large enterprises to a self-owned AI assistant that serves every individual. The focus of our AI is on each individual's value and the emphasis that everyone has the right to use customized AI. Therefore, we hope to develop a modular and combinable personalized AI engine.

                              The whitepaper is still in the process of evolving, we are always open for critics and opinion about anything about branche, also there are possible errors within. If you find any errors or inaccurate wordings, feel free to contact us and help us improve.


                              © 2023 Branche. All rights reserved.