Setting the file. One moment.
Skill 29 · Knowledge Base For Startups
Subchapter 29.91
references/learn/genai/blockchain-and-generative-ai-fueling-innovation-within-the-digital-economy.mdMarkdown16 KBView on GitHub
Generative AI is unlocking new possibilities for the onchain economy
Right now, powerful forces are coming together to create immense innovation and economic potential: the onchain (opens in a new tab) economy and generative artificial intelligence (AI). The onchain economy represents cryptocurrency transactions and activity on the blockchain, and introduces a new way to create, own, and exchange digital assets and data.
When these technologies intersect, institutions and startup companies across both traditional and decentralized industries can expect radical changes to how information is managed and monetized. Today, generative AI and blockchain technology are already fueling mutual growth across multiple industries. To prove what’s possible in this exciting new territory, we’re here to discuss how AWS is driving the convergence, including real-world examples and proven strategies to help you lead the way.
The onchain economy represents a new era of the internet, positioned to enhance user engagement, improve privacy, and democratize communication. Generative AI is now playing a pivotal role in unlocking the full potential of onchain technologies and assets across the blockchain stack. By integrating AI with blockchain technology, platforms and applications can create more secure environments, thereby enhancing trust and fostering the adoption of decentralized applications.
One significant application is the use of AI agents to analyze smart contracts for vulnerabilities, like reentrancy attacks and logic errors. By continuously monitoring contracts, AI agents can alert developers to potential security threats in real-time so they can respond rapidly. These agents can also scrutinize transaction patterns on blockchain networks to identify suspicious activities that could indicate things like fraud or money laundering. By profiling wallet behaviors, they can detect anomalies that suggest compromised accounts or malicious actors.
For developers, generative AI agents are supporting the adoption of blockchain technology among leading institutions and even startups, as they can autonomously transact and manage digital assets on behalf of users. Traditionally, users could only transact through peer-to-peer interaction, however, innovation across the intersection of blockchain and AI has created an opportunity for AI bots to transact on behalf of users. The use of AI agents introduces a digital future where agents can be used to mitigate risks, enhance automation, and authenticate content on the blockchain. The synergistic capabilities of these two technologies are uniquely positioned to bring the onchain economy to the forefront of innovation for both institutions and small businesses.
From the very start of AI development, preventing manipulation and misinformation is crucial. AWS offers a robust suite of services and tools to support responsible AI. For instance, Model Evaluation on Amazon Bedrock (opens in a new tab) allows customers to assess, compare, and select the best foundation models (FMs) for their needs, focusing on metrics like accuracy and safety. Additionally, Guardrails in Amazon Bedrock (opens in a new tab) enables users to implement safeguards that align with their responsible AI policies, automatically detecting and preventing queries and responses in restricted categories.
Blockchain technology has a decentralized framework that enhances trust by allowing users to fine-tune foundation models. This collaborative approach enables both proprietary and open-source models to learn from each other, ensuring checks and balances to prevent unexpected outcomes. AWS customers are leveraging this technology to address trust, transparency, and privacy concerns associated with generative AI. The immutable nature of blockchain ledgers is particularly powerful, as it allows for tracing and certifying the origins of digital assets, which is crucial in an era where AI-generated content blurs the lines of originality.
Blockchain technology and decentralized finance (DeFi) are paving the way for a more data-rich and powerful future for AI that benefits both large-scale institutions and startup ventures. While current models are trained on limited datasets, blockchain technology enables communities to sell or rent proprietary or private data, enhancing model learning and content generation.
Cryptocurrency is the backbone of the onchain economy, powering borderless transactions across geographical and political boundaries. Coinbase (opens in a new tab), a leading company in enabling wider use of blockchain-based digital assets and decentralized platforms, is at the forefront of this financial transformation. By marrying blockchain technology with generative AI, the business has created industry-leading conversational chat solutions for millions of users engaging with crypto assets.
Boosting productivity was key to Coinbase’s mission to increase economic freedom for over one billion people. However, they needed an agentic workflow, (a process that enables LLMs to reason and take multiple steps to complete complex workflows.) By using Amazon Bedrock (opens in a new tab)‘s simple APIs, Coinbase was able to quickly create these workflows for the digital assistant and auto scale capacity with spikes in utilization. With these capabilities, they can create personalized experiences whereby the chatbot quickly responds to customer queries with accurate account information. This also helped reduce handling times and improve the customer experience (CX) while enabling Coinbase to save costs.
With the help of the LLM Anthropic Claude 3 Sonnet (opens in a new tab), the platform can now handle complex reasoning with responses capable of following the principles and tone of voice of a Coinbase CX agent. By harnessing the faster, more cost-effective Claude 3 Haiku (opens in a new tab), it can also take on focused tasks like ensuring customer input meets the topic and content guardrails of the chatbot.
And it’s not just customers benefiting from Amazon Bedrock’s large language models (LLMs). When combined with the machine learning (ML) service Amazon Textract (opens in a new tab), employees can use LLMs to process complex documents and run natural language queries on them. Generative AI has unlocked multiple other internal use cases, including helping convert text to structured query language (SQL), allowing non-data scientists to query and obtain business critical data. Text-to-SQL conversion offers several other benefits, such as keeping sensitive data secure and compliant and supporting scalability while reducing latency. Having made huge breakthroughs, the business continues to collaborate with AWS to explore the opportunities within this emerging technology.
Managing and monitoring AI systems is a tough, multi-faceted process. Because of this, many enterprises are left open to risks like data privacy breaches and unintended model outputs. In the wake of trust concerns and new regulations, effective governance tools are a must-have for boosting AI’s transparency and auditability.
Leading the charge, Prove AI’s (opens in a new tab) software-as-a-service (SaaS) offering is a first of its kind system for tamper-proof AI governance. As a Web3 company, security and cryptography are at the heart of Prove AI’s technology. Their highly secure blockchain allows them to track data provenance and lineage end-to-end, in turn, providing visibility and control over AI systems. It’s an innovative approach that has been recognized by the National Institute of Standards and Technology (NIST) as an effective approach to generative AI risk management.
To make Prove AI possible, they needed a hybrid blockchain architecture to securely log and store a range of data, including everything from AI training datasets to model metadata, model context details, generative AI prompt session information, and related critical machine learning data. Already hosted on AWS, the business also chose to leverage AWS Key Management Service (KMS) (opens in a new tab) to easily generate key pairs for securely signing transactions onto both private instances and public blockchain networks. Prove AI was built to navigate different public blockchain architectures and securely manage AI data on top of Web3 technology.
The seamless integration of AWS KMS meant they could dramatically speed up time to market and bring down costs—they saved more than nine months in development costs versus building and integrating their own KMS, which would have been a 5x cost increase in ongoing key management costs. AWS has also helped Prove AI offer comprehensive intellectual property (IP) protection, authentication, and accountability for developing and deploying AI models. Datasets with copyright information can be tracked and traced back to their source, and cryptographically verified in case of IP infringement, a process that’s crucial for effective AI governance and compliance. By teaming up with AWS, Prove AI’s blockchain technology is making strides in its mission to derisk AI implementation and drive responsible adoption.
Thousands of companies, from large-scale platforms like Coinbase to startup ventures, are already using AWS for blockchain-related tasks: supporting their nodes, to securing keys, and managing their data analytics. Beyond AWS Key Management Service (AWS KMS), enterprises are breaking new ground with the help of fully managed infrastructure and data services like Amazon Managed Blockchain (opens in a new tab), a one-click self-hosted node deployment with Node Runners (opens in a new tab), and other key management services such as Nitro Enclaves (opens in a new tab).
Blockchain-native AI companies can further take advantage of AWS’s comprehensive generative AI stack comprising three key layers: infrastructure, tools, and applications. Infrastructure includes NVIDIA GPUs and custom silicon chips (such as Trainium (opens in a new tab) and Inferentia (opens in a new tab)) for foundation model (FM) training and inference. Tools such as Amazon Bedrock and Amazon SageMaker (opens in a new tab) help customers easily build with FMs and machine learning (ML) models. Applications such as Amazon Q (opens in a new tab) leverage FMs to accelerate software development.
The startup data platform Allium (opens in a new tab) unites several AWS services to power their generative AI tool. The company democratizes blockchain data and derisks activity with a simple way to query convoluted data. By overlaying natural language processing on top of index data, users can ask questions in simple terms (e.g., “How do I monitor ETH transactions for a specific address?”). AWS-enabled AI then translates questions into precise SQL commands, eliminating the need for deep technical knowledge. As a result, users ranging from novice hobbyists all the way up to financial analysts can easily derive insights from blockchain data.
ZettaBlock (opens in a new tab) (KiteAI) is harnessing AWS’s robust infrastructure and services to democratize AI development and expand its possibilities. As well as hosting comprehensive blockchain data, ZettaBlock (Kite AI) has access to over 100 million unique domain-specific queries. This data can be used to fine-tune LLMs via Amazon Bedrock, in turn improving the accuracy, traceability, and security of models. Their soon-to-be-launched decentralized AI protocol will be a base protocol layer focused around AI data, AI model builders, AI infrastructure, and agents. Using ZettaBlock (Kite AI)’s data infrastructure, they will be able to track provenance, protect data privacy, and provide fair compensation back to the ZettaBlock (Kite AI)’s ecosystem. Using Amazon Bedrock, they can also incentivize developers to contribute valuable AI inputs—thanks to blockchain monetization rails that can be easily integrated with their new protocol.
The opportunities for converging these technologies are abundant and blockchain native companies are primed to scale AI’s potential, but they can’t simply switch on the benefits without the right technical foundations. Decentralized LLM protocols still need a place to fine-tune and train models, as well as running inference. Rather than relying on compute-heavy processes, blockchain companies can hit the ground running using simple API calls. AWS’s specialized machine learning infrastructure, serverless models, and tools make this possible at scale. For instance, by removing the undifferentiated heavy lifting involved in building and optimizing infrastructure, Amazon SageMaker HyperPod (opens in a new tab) reduces foundation model training time by 40%.
Security, privacy, and safety are other core considerations for blockchain companies looking to gain enterprise adoption. Sharing proprietary data isn’t an option, so these features need to be built in for off-chain components. Creating guardrails around what a model should and shouldn’t do and gaining visibility into what is used to train a model is critical for sustaining the future of generative AI and blockchain technologies. Ultimately, training models takes time and adopting a new technology on top of an emerging technology won’t happen overnight. By integrating AWS services thoughtfully from the beginning, blockchain-native AI companies will have an easier time tapping into enterprise workloads. With the right strategies and tools, startups can monetize innovations for the long-term.
At AWS, we are committed to helping businesses unlock the full potential of blockchain technology, generative AI, and the onchain economy. Companies can manage risks and seize the rewards with the help of a comprehensive suite of services and rich expertise. Whether you are looking to launch solutions faster or scale them, there is a wide variety of AWS programs and support dedicated to helping startups at every stage including service credits, technical support, go-to-market opportunities, and rich training resources.
Ready to shape the future of decentralized technologies and AI-powered experiences? To kickstart your journey, gain high-level architecture insights, and discover more ways to leverage AWS core products and AI tools, email the Web3 team at AWS (opens in a new tab) to book a technical assessment.
Brad Feinstein is a GTM leader who builds and scales businesses at the intersection of enterprise technology and startups, with a focus on cloud, AI, and Web3. He leads AWS’s North America Startup FinTech & Web3 team, partnering with top startups and venture capital firms to accelerate market adoption and drive co-sell success. His experience includes leading GTM partnerships for international startups at AWS, serving as Head of Global Business Development at ConsenSys, and founding two venture-backed companies. Brad also held senior roles at Amex & Capgemini. He blends founder-level execution with enterprise scale to deliver customer success.
James Burdon is a Senior Blockchain Specialist Solutions Architect at AWS, focused on helping Web3 startups. James has over 25 years of IT consultancy experience and has been helping startups running on AWS for over 7 1/2 years.
John Liu has 14 years of experience as a product executive and 10 years of experience as a portfolio manager. At AWS, John is a Principal Product Manager for Amazon Bedrock, the easiest way for customers to build generative AI applications. Previously, he was the Head of Product for AWS Web3 / Blockchain, a managed service that helps customers build quickly with blockchain technology.
Source