Key Insights

• ChatAndBuild is converting AI from a black-box service into a verifiable asset class: mintable, auditable, and tradeable Non-Fungible Agents (NFAs) under the BAP-578 standard.
• Founded by Christel Buchanan, who was previously at Twitter, Brandwatch, and Zilliqa, ChatAndBuild has quickly forged institutional partnerships with organizations like Anthropic, OpenAI, Google Cloud, BNB Chain, the University of Oxford, Singapore’s Lorong AI, and leading Web3 infrastructure providers.
• NFAs transform AI from a feature to an asset class. They allow AI to become ownable, portable, and auditable. Every agent run emits Proof-of-Prompt (PoP), a Merkle-anchored, privacy-preserving record of configuration and outputs registered under BAP-578.
• ChatAndBuild has passed the first BNB application proposal (BAP), BAP-578, which allows ChatAndBuild to create and support its Non-Fungible Agents (NFAs).
• Positioned at the intersection of AI, blockchain, and no-code, ChatAndBuild’s NFAs aim to provide a new paradigm of learning AI Agents that are upgrades to traditional NFTs. They aim to revamp the NFT market and tap into the growing AI space.

Introduction

Barriers to software creation remain daunting for the vast majority of the world’s population. Mastery of programming languages, architectural complexity, and access to developer tools have traditionally confined innovation to a small cohort of technically skilled individuals. As a result, most ideas, no matter how creative or potentially impactful, remain unrealized, locked away by the technical gatekeeping of software development.

ChatAndBuild was conceived to break down these barriers. The platform enables anyone, regardless of coding proficiency, to articulate ideas in plain language and see them transformed into working applications, games, or digital tools in real time. Using artificial intelligence, ChatAndBuild handles everything from code generation to user interface design and backend deployment, abstracting away technical complexity and making software creation a more intuitive experience for users at all levels.

ChatAndBuild introduces its main feature, which is the Non-Fungible Agents (NFAs): an ownable, verifiable, and evolving AI agent encoded as a digital asset under the BAP-578 standard. Users not only build and deploy software but also create, own, and trade AI agents that evolve over time. This marks a significant departure from the traditional “AI-as-a-Service” format from the early 2010s, shifting toward an “AI-as-an-Asset” future paradigm where agents can be minted, owned, and traded, and where their intelligence and history accrue value directly to their creators.

With ChatAndBuild, the protocol aims to allow anyone, regardless of technical background, to build not just an app but an economy of agents: programmable entities that learn, assist, and persist across platforms.

Website / X / Discord

Background

Christel Buchanan, the founder and CEO of ChatAndBuild, has long championed the idea that technology should be an equalizer, not a moat.. Her global career spans leadership roles in scaling X across Asia, the Middle East, and Africa, as well as leadership positions at Brandwatch and blockchain platform Zilliqa. Her time spent at these companies gave her a front-row view of how innovation remains concentrated among the technically elite.

The genesis of ChatAndBuild can be traced to Buchanan’s observation of persistent barriers in software development. The emergence of generative AI, which allowed rapid prototyping and reduced dependency on traditional engineering, inspired her to envision a platform where anyone could become a builder. She recognized that the real bottleneck for innovation had shifted from technical expertise to imagination, and set out to remove the remaining obstacles.

Development of ChatAndBuild began in early 2025. The platform’s public launch in May 2025 saw immediate adoption, with over 150,000 users worldwide signing up within months. This early growth was driven by a conversational interface that enabled users to chat their ideas into real products.

Technology

ChatAndBuild currently focuses on two main entities. The first is ChatAndBuild, a no-code AI tool that helps individuals build web applications, games, and tools through its prompting hub for users. The other entity is Non-Fungible Agents (NFA), a protocol and marketplace powered by BAP-578. Every AI agent is minted with a verifiable Proof-of-Prompt (PoP), recording its creation, behavior, and ownership onchain. NFAs introduce Agent KYC, enabling enterprises to verify agent provenance and individuals to trade agents with trust and transparency.

ChatAndBuild: AI-Powered No-Code Builder

The platform’s mobile-first design allows users to build and deploy applications directly from their phones or laptops, making it accessible to a global audience. Over 40 languages are supported, lowering language barriers and encouraging culturally diverse software creation. Deep integrations with Figma (via Anima), Stripe, Supabase, GitHub, and more than 50 DNS providers streamline every stage from ideation to deployment. This seamless workflow abstracts away the complexities of code dependencies, package management, and infrastructure, allowing non-coders to focus solely on creativity and utility. Its AI-powered real-time streaming of code and design lets users see their applications being built as they converse with the system, providing immediate feedback and iteration.

Together, these systems aim to define a new category: buildable and ownable intelligence, as the first bridge between everyday creators and the emerging agent economy.

Non-Fungible Agents (NFA) and the BAP-578 Token Standard

At the center of ChatAndBuild’s innovation is the concept of Non-Fungible Agents (NFAs). Unlike traditional NFTs, which are static digital collectibles, NFAs are autonomous AI agents that can be minted, owned, trained, and traded as persistent onchain assets. Each agent is designed to evolve over time, retaining memory, skills, and user-specific customizations, with the intention to grow its intelligence over time and, as a result, its value over time as well.

Each carries its own persona, memory, and PoP, a cryptographically anchored record of how it was created, configured, and evolved through use. Over time, these agents can accumulate skills, reputation, and verified history, allowing their intelligence to grow.

With this vision, ChatAndBuild aims to empower users and transform the AI-as-a-service model into AI-as-an-asset, where creators, not platforms, retain ownership of the intelligence they build.

BAP-578: The Trust and Interoperability Layer

BAP-578 is an open token standard authored by the ChatAndBuild team and proposed for the BNB Chain ecosystem. BAP-578 extends ERC-721 to support intelligent, agent-specific functions, defining a unified set of interfaces, proofs, and compliance rules that make agents portable, auditable, and interoperable across ecosystems.

BAP-578 extends ERC-721 to support agent-specific functionality and introduces several key advancements:

• Proof-of-Prompt(PoP) Registry: Each agent run produces a Merkle-rooted attestation, a lightweight, tamper-evident proof of its activity, anchoring provenance and reproducibility without exposing sensitive data
• Agent KYC Metadata: Optional verification fields enable creators, enterprises, or marketplaces to attach identity, jurisdiction, and compliance tags, aiming to create a verifiable trust layer for regulated AI use.
• Gas Optimization: Transactions involving NFAs are designed to be 80–90% more efficient than standard token operations.
• Cross-Chain Continuity: Through integrations with tools like Hyperlane, NFAs maintain persistent identity and accumulated intelligence across multiple blockchains, ensuring portability and avoiding vendor lock-in.
• Dual-Path Architecture: Developers can deploy simple JSON-based agents for immediate, low-cost use or learning-enabled agents with Merkle Tree Learning for cryptographically verifiable intelligence progression.
• Merkle Tree Learning: Ensures that agent intelligence evolves in a tamper-proof way. Interaction and training data are organized into a hierarchical tree, with only the 32-byte Merkle root stored onchain. Updates require valid Merkle proofs, making each step of an agent’s learning history verifiable without inflating storage costs.
• Hybrid Storage Model: Onchain components secure identity, ownership, and Merkle roots, while offchain storage supports extended memory, personality data, and media assets, ensuring both security and scalability.
• Standardized Interfaces: BAP-578, establish consistent methods for action execution, state management, and logic upgrades, enabling interoperability between agents across platforms.
• Agent Templates: When first minted, agents will have two sets of templates: simple agents and Learning agents. Simple agents will be lightweight and less customizable, while Learning agents will have the entire capability of the BAP-578 upgrade.

Through BAP-578, ChatAndBuild aims to create the protocol layer for the Agent Economy, as a universal language for AI ownership, verifiable intelligence, and compliant agent exchange.

Security Framework For NFAs

As dynamic, intelligent assets, NFAs demand a security architecture that is more robust than static safeguards on NFTs. The ChatAndBuild framework introduces a three-tiered protection model that combines execution control, permissioned data governance, and cryptographic verification to preserve both trust and autonomy across the NFA ecosystem.

Layered Circuit Breaker Architecture

At the core is a layered-level circuit breaker system that allows risks to be contained at different scopes. At the blockchain level, governance retains the ability to enact a global pause in the event of systemic vulnerabilities or exploits. This ensures rapid containment of threats across all deployed agents and contracts..

At the contract level, developers and auditors can suspend a specific logic contract if flaws are discovered. At the agent level, individual NFAs have full autonomy to pause, unpause, or retire their own agents without affecting the broader system. Ownership rights remain sovereign, allowing users to protect their assets without triggering system-wide interruptions.

Permissioned Vaults for offchain intelligence

Each NFA includes an offchain vault, a creator-controlled environment where extended intelligence, such as memory logs, embeddings, or personality states, is stored.

Access to these vaults is cryptographically permissioned, with roles ranging from read-only analytics to full update control. Operations, such as model fine-tuning or memory rewrites, require signature-verified authorization by the agent’s creator or verified enterprise operator.

This design aims to ensure that while NFAs can evolve and personalize over time, no unauthorized actor can alter an agent’s intelligence or behavioral history. Changes are traceable, auditable, and provable against the agent’s PoP registry.

Proof-Anchored Integrity

Each action performed by an NFA, including interactions, configuration updates, or deployments, produces a PoP hash, a cryptographically signed record that captures system prompts, user inputs, model parameters, and outputs.

Only the Merkle root of each interaction set is stored onchain to maintain scalability, while detailed data remains encrypted within the agent’s vault and can be selectively disclosed for compliance or verification. This structure allows validators or marketplaces to verify claims about an agent’s configuration or performance without accessing underlying data. Together, these mechanisms replace mutable learning trees with immutable proofs of activity and authorship, improving auditability and verifiable history for agent operations.

Minting NFA Agents

For NFA agents, users will be able to mint two kinds of agents: simple agents and learning agents. Simple agents will be focused on providing an entry-level, interactive digital personality with a verifiable identity. Learning agents will aim to be enterprise-grade or creator-driven agents with evolving behavior and extended capabilities compared to the simple agent. Users will decide which agent they wish to mint, and that will ultimately determine how their agent operates.

Simple Agents: 

• Image/Video identity
• Audio voice file
• Persona Schema
• Light memory string

Learning Agents:

• All of the simple agent features
• Learning tree root (32 Byte)
• Learning module address
• Verifiable learning metrics

When a user mints an agent, they can choose the lightweight JSON configuration or mint a learning agent. With the lightweight agent, users will be able to pick attributes such as personality, and once minted, they will be able to interact with them. A big component for the simple agents is that while they are easy to make, they do not have the learning capabilities of a learning agent, and so do not have attributes that are changing and remain static, but they are able to interact with users on a level that is more than a typical ERC-721 NFT.

The simple agents will have personalities and can be interacted with. When minting a learning agent, it enables a learning functionality backed by Merkle proofs. In this model, every interaction or training update generates new data points such as preferences, outcomes, or confidence scores, which are organized into a hierarchical Merkle tree. Only the 32-byte Merkle root is recorded onchain, while the detailed learning tree is kept in the agent’s offchain vault. Any claim about an agent’s knowledge or capabilities can be validated against the root using Merkle proofs, ensuring the learning history is tamper-proof without burdening the chain with large data. This system enables agents to evolve transparently and cost-effectively, a critical feature that supports NFAs as appreciating cognitive assets.

Ecosystem

ChatAndBuild Ecosystem

ChatAndBuild’s ecosystem strategy emphasizes cross-sector and global collaboration, linking universities, enterprise infrastructure providers, and builder communities into a shared development network. Current partnerships include:

• BNB Chain: Primary deployment chain for BAP-578, supporting gas-optimized contracts and treasury integration.
• OpenAI: Used in ChatAndBuild for users developing their apps.
• Anthropic: Claude Sonnet 4.5 powers cross-agent orchestration and contextual reasoning.
• Google Cloud: Awarded Google Cloud grant to support infrastructure scalability.
• thirdweb: Provides SDK and relayer stack enabling low-friction minting and marketplace operations.
• University of Oxford (Jesus College): Hosts hackathons and academic collaborations across undergraduate and postgraduate programs.
• Lorong AI
• OpenSourced
• WomenInAI

Codechella

Codechella is ChatAndBuild’s global builder and community initiative that combines technical education, hackathons, and cultural events to engage developers, students, and creators. The 2025 Singapore Summit gathered over 200 founders, investors, and researchers from organizations including OpenAI, BNB Chain, and Babylon Labs. The Codechella Tour now spans six universities across Asia and the UK, fostering early adoption among students and researchers.

Codechella - The Cultural & Builder Flywheel

Codechella has become the cultural nucleus of ChatAndBuild’s community strategy, blending festival energy with technical rigor:

• Codechella Summit Singapore 2025 gathered 200+ founders, investors, and researchers (OpenAI, BNB Chain, Strategy, Babylon Labs and more).
  
• The Codechella Tour now spans six universities across Singapore, Malaysia, Vietnam, Indonesia, Korea, and the UK, culminating in national finals in November 2025.
  
• Specialized tracks like Codechella Kids and VIP Masterclasses (launched with Bytedance) turn family creativity and academic excellence into real product launches

Each node in this ecosystem amplifies another:

• Universities validate accessibility and generate talent.
  
• Governments provide regulatory legitimacy.
  
• Enterprises anchor infrastructure and capital.
  
• Communities drive cultural momentum and adoption.
  

Together, they form the proof-based, compliant, and globally inclusive foundation of ChatAndBuild’s vision: to be the AI Super App and standards layer powering the world’s next 8 billion creators.

Tokenomics

While the NFA ecosystem does not currently operate with a token, it has outlined a planned framework to monetize its learning agents and provide value back to itself. The model is designed to ensure sustainable revenue generation, align incentives across stakeholders, and create long-term value capture within the ChatAndBuild ecosystem.

Revenue Streams

• Minting Fees: Each agent minted carries a fixed fee of 0.01 BNB, providing predictable, scalable revenue that grows with adoption.

Revenue Distribution

• 60% to the NFA/ChatAndBuild Foundation for research, development, and grants that expand ecosystem capabilities.
• 25% to the Community Treasury to fund developer incentives, partnerships, and growth initiatives, particularly around the cognitive derivatives market.
• 15% to the NFA Staking Reward Pool to incentivize long-term participation and create deflationary pressure on circulating supply.

Roadmap

Near-Term (0–6 months):

ChatAndBuild’s immediate priorities focus on global rollout, expanded language support, and enhancements to NFA functionality such as advanced training, user customization, and improved onchain learning metrics.

A central milestone in this phase is the NFA Digits auction, which will introduce 100,999 unique agents to the market. This auction will be the beginning of the NFA product and will be the first entrance of these new kind of agents onto the BNB blockchain.

Medium-Term (6–18 months):

The expansion phase extends NFAs across multiple blockchain ecosystems. Cross-chain compatibility will be achieved through Hyperlane integration and the deployment of the BAP-578 on Solana and Ethereum. Development of cross-chain bridges will enable agent portability, preserving identity and intelligence while preventing vendor lock-in.

Marketplace upgrades will introduce advanced discovery, search, filtering, portfolio management, and early cognitive derivative instruments such as futures and options. ChatAndBuild will focus its growth efforts on creating more hackathons, academic partnerships, and community-driven initiatives.

Long-Term (18+ months):

The long-term outlook for ChatAndBuild is focused on further scaling and mainstream adoption. ChatAndBuild will look to position NFAs as a mature asset class while trying to introduce NFAs to Fortune 500 companies. At the financial layer, the introduction of derivatives such as futures, options, ETFs, and insurance could enable institutional participation and risk management. Longer term, this phase anticipates autonomous agent-to-agent economies, where NFAs transact and collaborate independently as artificial economic actors.

Closing Summary

ChatAndBuild sits at the intersection of artificial intelligence, no-code development, and blockchain infrastructure, providing a framework for creating, training, and owning intelligent agents as onchain assets. Through the BAP-578 standard, the project extends NFTs into verifiable, evolving entities known as NFAs, which carry identity, provenance, and learning records across networks.

Early milestones such as the NFA Digits auction, marketplace launch, and enterprise pilot programs will serve as initial indicators of adoption and usability. Over time, broader integrations, cross-chain functionality, and the potential emergence of cognitive derivatives could position NFAs as a distinct category of digital assets.

If this framework achieves sustained traction, ChatAndBuild may contribute to the development of an “AI ownership economy,” where the value created by intelligent systems is transparently verifiable and distributed across participants rather than centralized platforms.