Dialogue with XerpaAI BOB: Reconstructing Web3 Growth with AI Intelligent Agents, How to Break Through the Marketing Dilemma?

The imagination of Web3 is gradually being unleashed.

From August 25th to 26th, The Prince Park Tower in Tokyo, Japan, hosted a grand event in the Asian Web3 field - the WebX Tokyo Summit. As one of the largest Web3 conferences in Asia, this year's WebX was themed "Connecting the Nodes, Beyond the Screen". It brought together thousands of enterprises from around the world, over a hundred exhibitors, and many industry experts to discuss cutting - edge topics such as blockchain technology, digital asset compliance, and ecological innovation.

At the summit, former Japanese Prime Minister Taro Aso stated in his speech that Web3 and AI technologies are "the core of the new round of industrial revolution" and elevated the development of Web3 to the level of national strategy.

At this grand event, XerpaAI made its debut and launched the world's first "AI Growth Agent (AGA)", aiming to solve the core pain points in the growth of Web3 projects in an intelligent way.

The background for the birth of this product comes from the core problem of Web3 startups: growth. For Web3 startups, high marketing costs, difficult - to - scale community operations, fragmented user reach, and the traditional promotion model that overly relies on human resources and KOLs have always restricted their growth rate.

As a key component of the UXLINK ecosystem, XerpaAI took this opportunity to launch an AI - driven intelligent growth agency platform. It reconstructs the growth path of Web3 projects through automation, intelligence, and data - driven methods.

As an observer of emerging industries, 36kr also conducted an in - depth interview with Bob, the CTO of XerpaAI, on - site. They discussed its technological concept and ecological vision. Bob not only shared how XerpaAI uses AI technology to help small and medium - sized enterprises reduce costs and increase efficiency but also talked about topics such as how an AI agent that truly understands Web3, communities, and growth should accompany projects to success in 2025 when Agentic AI is on the rise.

The following is the transcript of the interview, edited by 36kr:

36kr: First, please introduce the founding background of XerpaAI. As a part of the UXLINK ecosystem, how do you define the "world's first AI Growth Agent"? What is its core mission? Additionally, what are the pain points of traditional growth methods in the Web3 field? How does XerpaAI propose new solutions with the help of AI?

BOB: XerpaAI was born in the UXLINK ecosystem. It originally stemmed from the real dilemmas we observed in the growth stage of Web3 startups, such as high - cost and low - efficiency manual marketing, over - reliance on KOLs but difficulty in quantifying the effects, and the fragmentation problem of cross - platform user acquisition.

As the first AI Growth Agent (AGA), our positioning is very clear: to be the intelligent growth partner of projects. Our mission is to help teams free human resources from repetitive tasks and shift to a more intelligent, automated, and scalable growth model.

Traditional growth methods often require a large budget. The technology industry alone spends hundreds of billions of dollars on user growth every year, but the effects vary. KOL cooperation is mostly based on experience matching and lacks data support. XerpaAI uses AI to generate high - quality content, automatically distributes it on platforms including X, Telegram, and TikTok, and accurately pushes it through a real network of over 100,000 KOCs and KOLs, helping projects triple their conversion rates while reducing costs by over 70%.

36kr: XerpaAI emphasizes the "intelligent growth engine". Does this mean it can completely replace human teams in the future? Considering the AI development trend in 2025, especially the maturity of autonomous agent technology, how does XerpaAI help startups transform from "manual" to "automatic"?

BOB: Our core concept is to build an "intelligent growth engine". It is not intended to replace human roles but to serve as a powerful assistant, helping teams focus more on strategy. Like an intelligent guide, XerpaAI can autonomously complete tasks such as user behavior analysis, incentive distribution, and promotion strategy adjustment, while humans can focus on long - term product and ecological construction. After 2025, AI autonomous agents will become more and more mature, and XerpaAI is evolving in this direction - making growth truly "self - driven".

36kr: How do you combine AI technology with Web3 native elements? Can you briefly explain how XerpaAI works?

BOB: XerpaAI's technical architecture is a highly modular multi - AI Agents system, specifically designed to handle complex tasks in Web3 growth, such as automated user acquisition, community expansion, and KOL/KOC matching. We built the entire system as a collaborative agent network, where each agent focuses on a specific subtask and achieves seamless collaboration through shared states and communication protocols (such as blockchain - based smart contract verification). This is a multi - agent intelligent workflow (agentic workflows), where each agent can autonomously plan, execute, and optimize action paths, thus realizing an end - to - end intelligent growth engine.

Photo provided by the enterprise

Specifically, XerpaAI's architecture revolves around a central AGA (AI Growth Agent) coordinator, which is responsible for supervising the interaction of multiple dedicated agents, forming a dynamic decision - making structure. The agent network includes:

Planning Agent: As the entry point, it is responsible for decomposing high - level growth goals (such as "increasing the user conversion rate of a certain DeFi project") into executable subtasks. It uses the "Plan - and - Solve" prompting strategy, an advanced zero - shot reasoning method that can systematically formulate comprehensive plans to gradually solve tasks such as content generation, KOL matching, and performance optimization, ensuring that no key steps are missed.

Data Collection Agent: It collects and pre - processes multi - source data from the Web3 ecosystem in real - time, including blockchain transactions, social graphs, and cross - platform user interactions, providing real - time, structured data input for other agents.

Content Generation Agent: It focuses on generating multi - language and multi - modal content (text, images, videos, etc.). With the help of the "Zero - Shot Chain - of - Thought" prompting, it can generate high - quality content without pre - trained samples.

Distribution & Matching Agent: It is responsible for intelligent matching and content distribution, integrating Web3 native elements such as social graph analysis and link - to - earn mechanisms, and optimizing task execution paths through multi - agent collaboration.

Optimization & Feedback Agent: It monitors key indicators such as conversion rates and costs in real - time and continuously adjusts strategies through self - reflection cycles, significantly improving effects and controlling costs.

Integration Agent: As a bridge between AI and Web3 components, it supports decentralized verification and adapts to the needs of different sectors, such as DeFi liquidity incentives and SocialFi community building.

The communication between agents is achieved through a shared knowledge graph based on GraphRAG technology, supporting real - time data ingestion and reasoning. The system also introduces a "Memories" mechanism, storing historical tasks, user preferences, and optimization results through long - term context, enabling agents to continuously learn and self - improve. For example, if a promotion campaign fails to meet expectations due to insufficient incentives, the system will record the reason and automatically adjust the incentive mechanism in future strategies.

Overall, through the efficient collaboration of multi - AI Agents and the application of advanced prompting technologies, XerpaAI has automated and intelligentized Web3 growth tasks and has served over 110,000 communities so far.

36kr: In the AI breakthroughs of 2025, small specialized models and inference time calculation are becoming the focus. Does XerpaAI use similar technologies to process massive data? How does its data analysis engine ensure real - time feedback and self - optimization?

BOB: Yes, XerpaAI widely uses small specialized models to efficiently handle specific tasks, such as KOL matching and cross - platform content distribution. These models are optimized for Web3 data types, significantly reducing inference time while maintaining accuracy, meeting the requirements of scenarios with high real - time demands.

Combined with the development trend of inference time calculation in 2025, our engine uses efficient algorithms to process large - scale data. For example, the system can screen the optimal matching objects from over 100,000 KOLs in real - time and coordinate distribution on platforms such as X, Telegram, and TikTok. The data analysis engine achieves self - optimization through continuous machine - learning cycles: it collects user interaction data and applies reinforcement learning techniques to dynamically adjust distribution strategies and content generation methods, avoiding overfitting while continuously improving promotion effects and adapting to changes in different communities.

Photo provided by the enterprise

36kr: Currently, XerpaAI has served over 110,000 communities. How do you use multi - modal AI to automate user acquisition and community interaction?

BOB: We use multi - modal AI to analyze text, images, and even short - video content, automatically generating promotion materials suitable for users on different platforms and in different languages. For example, the system can generate interactive topics in real - time based on community discussion hotspots or customize exclusive content for KOLs. All of this is based on strong data - processing capabilities, allowing us to achieve large - scale interactions on platforms such as Telegram and X while maintaining the relevance and attractiveness of the content.

36kr: XerpaAI has over 100,000 KOL and KOC resources. How does AI help these content creators increase their influence and monetization efficiency?

BOB: Each influencer receives personalized support from XerpaAI. The system can automatically generate more suitable content suggestions based on their fan profiles and dynamically adjust incentive strategies to help increase their income. For example, when a KOL promotes a DeFi project, the AI not only provides copywriting but also recommends the best posting time and interaction strategy. As a result, their monetization efficiency is improved, and they are more willing to cooperate in the long - term, while project parties also gain more real and extensive distribution channels.

36kr: During the cooperation process, how do you ensure the data privacy of KOLs and users? How do you ensure the fairness and transparency of the incentive mechanism?

BOB: We use blockchain and smart contracts for the automatic execution and verification of incentives. Every share is transparent and tamper - proof. In terms of data, we use technologies such as decentralized storage and zero - knowledge proof to protect privacy. All of these allow us to operate efficiently while building a foundation of trust with KOLs and users.

36kr: XerpaAI has just received a $6 million seed - round financing. What are your future development plans? Can you give an example of how you helped a Web3 startup grow from scratch?

BOB: This funding will be used for product R & D, team expansion, and global market layout. Previously, we helped a newly - launched Web3 social project. Through AI - generated content and precise KOL matching and dissemination, we built a community of 100,000 people within one month, and the activity increased by two times. From content creation to distribution, from user connection to incentives, all were automatically completed by XerpaAI.

Photo provided by the enterprise

36kr: As the technical leader, what are you most looking forward to in "AI + Web3"? Finally, do you have any words for potential partners and users?

BOB: I look forward to AI truly promoting the maturity of Web3 and realizing a self - optimizing decentralized ecosystem. XerpaAI hopes to help more excellent teams overcome growth bottlenecks, quickly connect with users, and expand the market. If you are looking for a more intelligent growth method, welcome to visit our official website for more information, and we also welcome exchanges and cooperation at any time.