Update from the COO

2023 has been a year of massive progress, growth, and expansion for SingularityNET and our ecosystem. The details are contained below in this report, but at a high level, I am happy to report on a very successful year with multiple highlights. We have:

• Materially progressed in our core mission of advancing towards human-level AGI;
• Increased focus on our platform technologies with a massive shoring up of resources on the platform;
• Strengthened the SingularityNET foundation leadership team with some key new executive hires;
• Made significant strides towards progressively decentralizing the foundation and its governance with a newly elected Supervisory Council;
• Funded 60 projects building AI on the SingularityNET platform through our Deep Funding programme;
• Forged multiple new impactful partnerships;
• Materially expanded our marketing reach and presence;
• And last — but absolutely not least — successfully launched Rejuve Network, Cogito, HyperCycle and SophiaVerse into flourishing and fully independent entities in the true spirit of decentralization, as well as launching TrueAGI as an independent business.

SingularityNET is undoubtedly one of the most exciting and highest performing projects on the planet right now! My huge and heartfelt gratitude to all of you in the community who support all of our endeavors — please stay involved, keep voting, listening, reaching out and growing with us on this incredible journey.

Janet Adams, January 2024

SingularityNET 2023 Highlights

As part of our organizational restructuring executed at the end of 2022, we reduced our operating costs by nearly one-third going into 2023 without compromising core technology delivery or support for incubated projects. Though a difficult decision requiring parting ways with valued team members who played a key role in our successes throughout the years, it allowed us to enter 2023 more nimble, focused, and positioned for sustainable growth. The phenomenal success of ChatGPT in January 2023 positively altered the financial landscape and enabled us to build and expand our technology capability.

Led by CTO Sergey Shalyapin, the new platform and blockchain team accelerated the delivery of key decentralized infrastructure components and made significant progress in our Cardano migration process. We have streamlined our platform onboarding process, modernized our AI publisher portal, launched our hybrid AGIX staking solution on Cardano, and initiated integrations with Knowledge Graphs and LLMs.

Our platform development strategy is closely aligned with our AGI strategy, building a bridge from decentralized AI now to a future of decentralized AGI. We have deepened our technical expertise, strengthened existing relationships, and forged new, productive partnerships with industry-leading players, such as Dfinity, Fetch.ai, and Vechain. These advancements pave the way for the deployment of new AI services, starting with the highly anticipated Deep Funding-powered services. Get ready to see the marketplace evolve even further with the introduction of the SingularityNET Platform Assistant and more advanced features just around the corner.

Beyond restructuring our platform and blockchain development team, we bolstered our ranks with key leadership hires. Alex Blagirev joined as Strategic Initiatives Officer, and Loic Claveau took the helm as Chief Marketing Officer. This leadership infusion was complemented by onboarding a team of value engineers dedicated to partnership success, and experienced marketers focused on brand building and increasing ecosystem visibility.

While we are focused on strengthening our position in the decentralized AI space, we continue to shape the technologies of tomorrow. Our AGI R&D initiatives advanced strongly throughout 2023, meeting our targets and incorporating emerging technologies. Building on the foundations laid by Dr. Ben Goertzel, Dr. Matt Iklé, and Dr. Alexey Potapov, our core AGI team has transitioned into execution and implementation, focusing on scaling and delivering the Alpha version of the OpenCog Hyperon AGI framework — chief components being the MeTTa interpreter and Distributed Atomspace (DAS). In 2023, we also partnered with Greg Meredith and his company F1R3FLY to dramatically accelerate the scaling of MeTTa through advanced Rholang concurrency methods.

Our commitment to creating decentralized, democratic, inclusive, and beneficial Artificial General Intelligence has never been more resolute. There is still more R&D needed to get to human-level AGI, but we have charted a clear development roadmap to scale Hyperon by incorporating numerous efficiency and scaling improvements across multiple spatial and temporal scales. Our roadmap also includes an array of AI algorithms and conceptual principles, grounded upon solid cognitive science foundations, all designed to work in concert with each other and facilitated by MeTTa and the DAS. To date, we have hit our milestones on schedule and met all of our commitments.

Our journey to AGI demands not just scientific brilliance, but collective wisdom and values that can guide the evolution of AI toward AGI systems that are both highly capable and beneficial for humanity and all sentient beings. This is why it is imperative that our progress on the decentralization front matches the acceleration of decentralized AGI software development.

Recognizing that AI and AGI stewardship should not be confined within SingularityNET’s walls, we are working tirelessly to empower you, our community, to provide the checks and balances, the voices of debate, and constructive dissent to increase the chances of creating true, beneficial intelligent systems at the human level and beyond. Concretely, we have started involving you in decisions relevant to the future of our Ecosystem by creating a Deep Funding Focus Group, expanding our Ambassador Program, hosting the first Decentralized Governance Summit, holding the second Supervisory Council Election and other community governance events, among other initiatives. The Beneficial AGI Summit in February will further power us forward in our uncompromising commitment to positive technological impact on humanity and the planet.

Throughout 2023, we took a series of concrete steps to execute our plans, expanding our ecosystem, and defining a clear strategy to become the Knowledge Layer of the Internet (Blockchain Layer 3) and lead in the era of decentralized AGI. Each of these moves opens new opportunities for our ecosystem, and their impact will continue to be felt far into the future.

Last but not least, we are grateful to everyone who supported our mission to decentralize the future of AI development by participating in our token generation events last year. Your support and passion are what made the Rejuve.AI, HyperCycle, Cogito Protocol, and Sophiaverse projects a reality. As these projects mature, we will continue to onboard their AI agents on the SingularityNET marketplace, powering us forward to our goal of decentralized beneficial Artificial General Intelligence and providing useful AI for worldwide benefit.

Pod and Project Overviews — In their own words

SingularityNET Organization Updates — In their own words

OpenCog Hyperon

During Q1, Minecraft AI mod Vereya was developed, extended, and published, with our AI agents keeping up-to-date with new Minecraft versions and working on different platforms. This allowed us to compare our AI agents to the GPT4-based Voyager agent.

In Q2, our MeTTa interpreter became ready for internal testing and experimentation, with improvements related to variable bindings, type checking, expression indexing, and other features. A Space API was implemented and improved, and examples of an SQL space and a “neural space” with its own pattern matching via LLM calls were provided, opening possibilities for further integration with other types of spaces.

Further improvements of MeTTa were made during Q3, including code documentation and an advanced REPL facilitated a burst of AI algorithms and Hyperon components prototyping in MeTTa, including Probabilistic Logic Networks PoC, Non-Axiomatic Reasoning System (NARS) implementation in MeTTa, SingularityNET dialog system port to MeTTa, experiments with Pattern Mining, and others.

In the last quarter of 2023, the first version of the Distributed AtomSpace integration to MeTTa using Space API was completed. This was done together with considerable progress in the DAS development, including its caching and indexing components, APIs, and its application to Bio-Atomspace.

• A MeTTa module for LLMs integration has been prototyped as a basis for future replacement of the SingularityNET dialog system, and the first version of the SingularityNET Platform and Marketplace Assistant has been implemented within it and run as a telegram bot for initial testing.
• A Minimal MeTTa core component has been developed, which allowed rewriting the Rust implementation of MeTTa interpreter using the minimal set of core operations, making MeTTa even more flexible and amenable to metacomputations and inference control.

For a high-level background & introduction, see OpenCog Hyperon: A Framework for AGI at the Human Level and Beyond, published in October 2023.

Outlook for 2024 —
The most important milestone for Hyperon in H1 of 2024 will be the Alpha Release, which will include a MeTTa packages management and deployment system and useful packages and extensions themselves, tutorials, examples of Hyperon applications to practical use cases involving DAS, and AI libraries in MeTTa. Post Alpha Release plans will focus heavily on AGI R&D and the related expansion of Hyperon components. An important milestone will be achieving real-world interoperability of Hyperon and the SingularityNET platform. This will include the platform SDK in MeTTa allowing the use of services as Hyperon components, implementation of AI-DSL as a part of MeTTa SDK allowing its utilization both in services and clients with gradual development and adoption, further improvement of the platform assistant and its integration with MeTTa SDK.

F1R3FLY Update

In 2023, SingularityNET partnered with Greg Meredith and his company F1R3FLY to dramatically accelerate the scaling of MeTTa through advanced Rholang concurrency methods, and we are marching steadily towards demonstrating concurrent MeTTa execution. The demo is designed to show a near-linear scale-up of execution as we add hardware and nodes to the network. Towards this end, we have:

• Ensured the now completed (first iteration) of the Rust implementation of RSpace works well in the network;
• Developed example MeTTa programs for demonstration, and are reviewing MeTTa programs under development by the PLN team as candidates for the demo;
• Refactored the MeTTa to Rholang compiler (HeavyMeTTaL) to support debugging and language modifications;
• Engaged Oracle as a channel partner to help us deliver this demo by providing the cloud infrastructure to support testing, staging, and production deployment.

Additionally, we have been developing an extension of the Operational Semantics in Logical From algorithm (OSLF) to account for a realizability semantics for PLN. This means that PLN formulae are interpreted as collections of MeTTa programs. This makes for a very powerful query model. Using OSLF means we have algorithmic means to generate both the formulae and the semantics/query model.

OSLF is joint work with Mike Stay. We are also working with Ornela Dardha and the University of Glasgow. As part of outreach and development for OSLF, F1R3FLY.io gave a talk at Glasgow’s PLUG seminar in December. We are also collaborating with Jonathan Warrell on the probabilistic extensions to OSLF.

In the spring of 2023, we published a paper on an operational semantics for MeTTa, which we submitted as the basis for

• a “yellow paper” in the style of Ethereum’s yellow paper;
• and the basis for judging the correctness of the HeavyMeTTaL compiler.

In the Fall, we updated that approach, producing 4 different calculi for MeTTa. We believe a calculus is absolutely essential for adoption by the programming language semantics community. We also believe that the Rholang calculus can help answer several outstanding questions regarding MeTTa semantics.

For bonus points — F1R3FLY.io developed an algorithm to compile rholang and other programming languages into hypervectors. We are reducing this to practice. Here is the very beginnings of the compiler. F1R3FLY.io has engaged the compiler team at GSI who have developed a version of a hypervector library for the GSI APU Gemini processor line. Current performance estimates show that this method of execution will be 1000x faster.

AI-DSL

The AI-DSL project roadmap underwent considerable change over the course of the year due to the advent of new technologies and tools.

During Q1, we built prototype tools in Idris to crawl the SingularityNET marketplace, retrieve relevant protobuffer information about the SingularityNET service, and verify types and subtypes.

Due to rapid MeTTa development, we quickly switched in Q2 from working with Idris to our original goal of developing MeTTa tools for AI-DSL

With the advent of systems such as ChatGPT, and the release of HuggingGPT as a tool for discovering and connecting services, we began a thorough investigation of HuggingGPT capabilities. We found that, while HuggingGPT was an impressive system, it was quite brittle, oftentimes breaking and not working as desired. As a result, we decided to fork AI-DSL into two separate tracks:

• The SingularityNET Platform Assistant, with a planned release date in H1 of 2024, aims to provide an easy-to-use system for interacting with the SingularityNET platform (for instance, requesting services or combined workflows in natural language via LLM);
• While also continuing the original AI-DSL work, focusing on core MeTTa SDK tools to enhance the reliability and robustness of the SingularityNET platform assistant.

Q3 saw substantial progress on required basic SDK processes such as chaining, typing, and sub-typing within MeTTa.

Q4 saw the build completion of protobuffer to MeTTa parser supporting nested definitions and methods and the incorporation of protobuffer to MeTTa parse into marketplace crawler.

Outlook for 2024 —
We will work closely with the MeTTa, SingularityNET Platform, and SingularityNET Platform Assistant development teams to facilitate requirements that align with the AI-DSL development goals of providing OSS developers with an easy-to-use onboarding experience, and marketplace users with a robust and simple tool to discover and connect services meeting their needs. When complete, the SingularityNET Platform Assistant front-end, with an AI-DSL back-end, will ultimately lead to a system that can AI agent autonomous actions and interactions, providing a unique AI environment that can rapidly accelerate platform growth through creating exponential utility for all services.

AI Platform

The AI, blockchain, and Platform development team provides technical support for all products, and product users (staking platforms, wallet linking tool, Publishing Portal, etc). We process requests for technical support promptly and help to cope with emerging difficulties.

2023 saw an overall code refinement and updates to utilize the latest available technologies. This includes fixing minor suboptimal components of the Ethereum staking portal and the Airdrop portal, and establishing Ledger support from the Cardano blockchain for the convenience of users. We provided support during the TGE of new tokens, configured the API to determine the circulation supply, and created snapshots to record the status of user balances for the further formation of white sheets within the framework of the processes described below. In addition, we are working on implementing WalletConnect on the staking portal for the convenience of users. Some of our biggest accomplishments and milestones include:

• Wallet Linking Tool — We have developed and implemented a service to create collections of user wallets. Currently, the service supports wallets from the Ethereum and Cardano blockchain networks. In the future, it is planned to add more wallets and networks to the service. As the service expands its capabilities and applications, it aims to become a single place for anonymised user identification. The service already allows companies to optimise and support participants of both networks during the voting process, taking into account the user’s balance of all wallets in the collection, as well as during the airdrops and distribution of tokens.
• Staking on Cardano — 2023 finally saw the launch of staking on Cardano. The solution is technically complex and required a lot of effort due to the characteristics of the Cardano blockchain and the need to create a high-quality and secure solution. Currently, staking on Cardano is presented on the SingularityNET portal with AGIX tokens, but in the future, it will be possible to stake with any token.
• SingularityNET Bridge — A major update of the front end of the SingularityNET bridge has been carried out, which has made it possible to correct the logic of the interaction with the wallets on the Cardano side. Now the interaction is fully automated and secure, the behavior duplicates the logic of working on the Ethereum side.
• Binance Support —Support for the Binance blockchain network was introduced. Transfers are now available between Ethereum and Cardano, as well as Ethereum and Binance Smart Chain. Future plans include adding support for transfers between Binance Smart Chain and Cardano and updating the bridge’s general front end to combine support for all networks within one page. In addition, Cogito’s CGV and Rejuve’s RJV tokens have been integrated into the bridge.
• CLI, SDK, Daemon — We have started development to automate the deployment of daemon modules (a core component of the interaction between an AI developers service and the Platform) and simplify the interaction with the SingularityNET CLI. We are moving towards a global onboarding methodology and multi-platform tools and simplifying the interaction with these tools. We are also working on separating the CLI from the SDK and developing two versions of the SDK: JavaScript and Python. Since we are rewriting the documentation at the same time as the development, the release and the actual documentation will be presented together. To make the work of service providers more convenient, a new way of interaction between the daemon and the service using HTTP requests has been implemented in the daemon. Previously, connection to the daemon was limited to grpc, json-rpc, process (threaded interaction with the running provider service). However, service providers now have options for deploying their services on hosting based on serverless GPU computing (for example, banana-dev). This functionality is currently being tested and debugged. Also, there were implemented new functionality for service providers to train models. This allows service providers, as part of interacting with the daemon, to activate an additional model training service based on the provided data dataset. The functionality is being tested and improved to meet technical requirements.
• Publisher Portal — We have initiated a comprehensive enhancement of the Publisher Portal, focusing on refining its logic and user interactions to ensure seamless and user-friendly onboarding processes. The testnet has undergone a transition, and we have meticulously refined the interaction logic with the daemon at the testnet level. Additionally, an intensive phase of enhancing the front-end functionalities of the Publisher Portal is underway, accompanied by thorough testing. This effort aims to address instances of unexpected behavior, ensuring a smooth and reliable experience for users interacting with the portal.
• Voting — We improved the functioning of the portal for voting, and also integrated a new service for identifying users by their collections of wallets. This made it possible to significantly simplify the voting process and, within one procedure, take into account the balance of all user addresses without the need to vote with your other wallets again. This also made it possible to take into account the votes of users using the Cardano blockchain network.
• Jam Galaxy Portal — The team developed the jam-galaxy portal. This advanced platform for artists and musicians combines various mechanics to support creative people. The platform includes several types of users and entities, such as fans, creators, and also “artists” which is the entity that unites the entire creative team — from songwriters and music writers to technical support and advertising specialists. Each “artist” owns his personal forum and public page. These initial features will be significantly expanded, and the Jam Galaxy project itself will share the details. The platform will support the ability to use AI services to process music, highlight specific musical instruments, generat music, and many other services from the SingularityNET platform and other resources. In addition, each Artist will have his own tokens, and the Artist’s fans will be able to receive them and other bonuses, as designed by the artist themself, and so much more.
• GCOIN — We have developed a demo of a new generation stable coin GCOIN. As part of the demo, users could get acquainted with the key mechanics and features of the price formation of the new stablecoin. Mint the token or burn it, and also monitor the dynamics of price formation and its changes in the case of the token itself. (The GCOIN project has pivoted, and is now being developed as a yield-bearing stablecoin under the current name ‘cgvUSDC’).

Finally, we’d like to share some updates on AI Service Hosting, a new feature we are developing for the Platform that will be of significant benefit to developers:

We are currently at the forefront of implementing cutting-edge event-driven solutions tailored for artificial intelligence (AI) models within the SingularityNET platform. This strategic endeavor not only represents a paradigm shift in AI model deployment but also carries profound implications for cost reduction and the optimization of scalability.

Traditionally, service providers bore the substantial burden of maintaining dedicated servers, incurring significant financial investments. The adoption of an event-driven service architecture solves this landscape, ensuring that computational instances are invoked only when required by end-users. This approach allows for the scalable provisioning of GPU models, enabling a flexible and efficient scaling from zero utilization to the predefined upper limits set by the provider and back. Our commitment extends beyond mere infrastructure optimization as well — we aim to diversify the serving model hosting and interaction methodologies.

This initiative includes incorporating modern operational techniques, such as leveraging operators within Kubernetes and other advanced AI cluster management systems. This diversification broadens the spectrum of hosting options, empowers service providers with enhanced AI prototyping capabilities, and expedites the seamless migration of models onto the SingularityNET platform. In a holistic context, adopting GPU-enabled serverless microservices within the SingularityNET platform yields many advantages. These encompass not only the evident reduction in operational expenses but also the amplification of scalability, heightened reliability, expanded options for model placement, and enriched model interaction.

AI Services released on the platform

The following 24 services have been added or updated to the AI platform during 2023. This increased the number of Platform services from 68 in January to 83 by the end of December 2023.

Deep Funding

Our community-driven Deep Funding (DF) initiative has grown significantly, especially in the last round. To make this growth sustainable, we need to focus on tools and processes.

From a tooling point o