陶芳波

The key dividing line repeatedly identified in this discussion is not demand, but whether unit economics can turn positive on every call. 朱哲清 estimates that 80%—90% of industry costs now sit in inference. A single Deep Research run exceeding 1M tokens costs roughly $3—4 on domestic models and $8—10 on top overseas models; Grok can run into the tens of dollars. He has “not seen a single company able to break even on costs” in a C-side use case. Survival routes include model routing, context compression, vertical fine-tuning and self-hosted serving. For C-side products, the question is also whether model providers can cut prices by roughly 80% while leaving room for profit; otherwise, continued cash burning eventually ends in collapse.

Cursor’s $10B valuation and the simultaneous backlash over limits on its $20 plan show that application-layer growth does not automatically become gross margin. 泓君 cited the company’s $900M financing round in June and the view that Cursor and GitHub together contributed roughly $1.2B of Anthropic’s approximately $4B revenue last year. 陈志博 put it bluntly: Cursor is “basically working for Anthropic.” With Claude, OpenAI, Gemini and xAI all treating coding as a primary battleground, Cursor must absorb upstream inference costs while competing head-on with its suppliers. 泓君 speculated that the pricing change may also be intended to improve gross margin ahead of the next financing round.

Poke AI is trying to turn the entire web’s toolset into the model’s action space, tackling tool discovery and context costs at the same time. A single model may support only tens of thousands to 1M tokens of context, while descriptions of all web tools could run to tens of millions of tokens. Traditional Agents replay the full history and tool information at every step, causing the same content to be processed 20, 30 or even 50 times. Poke wants developers to call an Agent as easily as they call the ChatGPT API, but model reliability still falls sharply when it faces thousands of highly similar tools.

The three companies are betting on three different moats: tool ecosystems, identity networks and design delivery. Second Me distinguishes between the “doing” of a profession and a person’s “being,” aiming to give everyone an identity Agent that represents them and connect those Agents into a network. Lovart is not competing with foundation models such as Nano Banana for point-generation capability; it uses a designer-style planner, aesthetic memory and an editable canvas to deliver personalized work. For Lovart, model capability is raw material: “Only water deep enough can carry a ship large enough.”

Training paradigms are shifting from SFT toward RFT and RL because of the data bottleneck, but “learning through use” remains far from solved. 朱哲清 believes the most expensive part of SFT is now licensed data and labeling for complex tasks. RFT/RLVR comes at the cost of longer, less stable training and weaker generalization. 陶芳波 summarized the shift as: “The SFT era was about finding data; the RFT era is about finding environments.” Their key disagreement is whether the next generation will enter an era of experience: 陶芳波 believes online use will make Agents smarter, while 朱哲清 stresses that few-shot feedback may be an outlier and that safe inference-time updates are “extremely difficult.”

B-side Agents are already close to breaking even, while C-side products are taking three different bets: subscriptions, network effects and outcome-based revenue sharing. Poke can cover costs on the B side; on the C side, subscriptions may cover inference but not labor. Lovart uses tiered credit subscriptions, testing delivery value through whether users keep buying. Second Me is choosing to charge as little as possible in order to build an identity network; the possibility of an AI-native advertising model is 泓君’s growth bet, and 陶芳波 agreed. Pine takes a cut of the amount it recovers in users’ billing disputes, offering the most direct answer: “Take a percentage of the final outcome.”

Trust in and the durability of high-priced KOL traffic are weakening, while community and the Agent kernel are becoming more important distribution and platform assets. Bill believes expensive KOL traffic decays quickly and carries weak trust; Runway and Luma had more durable early growth through co-creation on Discord. 陶芳波 sees the Claude Code SDK as a deeply underestimated “coding kernel” that could even become the CUDA of the Agent era. 泓君’s changed view of Cluely, Lovable and Replit also shows that value need not come from fully automated delivery: real demand can exist in live summarization, prototype communication and giving people more confidence.

🔗 Original source & video: E207 | Agent Founders’ Cost Breakthroughs and Commercialization

陶芳波 is betting not on a better AI assistant, but on a unique, trainable and distributable “identity foundation model” for every person. Jarvis-style him and Her-style her are external roles; Second Me aims to internalize memory, preferences, emotions and values as parameters, allowing AI to autonomously connect with services, people and other Agents as “me.” The platform-level vision is a structurally similar AI replica of human society, but with more efficient connection, negotiation and decision-making.

Mindverse’s technical dividing line is its refusal to fake personalization with one foundation model plus different RAG knowledge bases. 陶芳波 believes thinking happens in model parameters: if everyone shares the same model, “at bottom, there is no difference between me and you.” The team first rewrites raw data into first-person, subjective memories organized around people and issues, then fine-tunes an AI-native memory. At this stage, about 100 memories, equivalent to roughly 20 hours of data, are enough to get started; 7B is considered a suitable range, and daily training may cost less than $1.

Open source is the most important trust strategy for this kind of sensitive infrastructure. Second Me supports local training and deployment, so personal data need not go to the cloud, while the model can still connect to the internet. The project reached 10K GitHub Stars in about three weeks and roughly 12K by the time of the episode; more than 10 users even spent tens of thousands of yuan on computers to train themselves. The community defines it as a “middle layer for digital identity,” or even “the interface between people and the AI era,” showing that demand is driven not only by functional efficiency but also by the powerful psychological desire to preserve oneself.

The most immediately deployable products are not fully autonomous task execution, but expression, judgment and information compression. Me.bot’s Talks can generate 1—3-minute audio messages from personal memories for specific audiences, while allowing listeners to ask questions in real time. With only about 30 seconds spent writing a prompt and roughly two minutes of waiting, 陶芳波 completed a startup talk that would normally require 1—2 hours of preparation. “Resonance” lets two Second Me models match and break the ice in about 5—10 seconds after phones touch, or helps a record that is neither fully public nor fully private find genuinely relevant respondents.

The distribution strategy deliberately avoids the cold start of a two-sided network: people who receive Talks do not need to have their own Second Me first. Content can enter existing recruiting and social scenarios through H5 pages; over the longer term, a Second Me Server similar to MCP could provide personalized scoring, filtering and judgment to websites and services. 陶芳波 acknowledges that China’s internet is currently “closed off into islands,” with limited API access, but believes universal identity will eventually clash with closed data loops. Whoever becomes the neutral identity layer trusted by users could occupy the position of new infrastructure.

The near-term monetization answer is to charge individual users who rely heavily on identity services; the longer-term vision is to charge service providers that call on identity. If users control their identity models, e-commerce, content platforms and advertising systems should theoretically pay identity owners for more accurate preference and decision data, with Mindverse taking a service fee. That would turn individuals from “platform products” into owners of their identity assets. 陶芳波 explicitly calls this “a beautiful idea” and says he is unsure it can be realized; the commercial loop remains unproven.

The main unresolved issues are data collection, real-time guardrails, platform access and user habits. WeChat, iPhone and other incumbents naturally possess more data, and large companies can enter the field; the startup window lies in having a “new species” first serve a small group of committed believers and then cross the chasm, rather than competing on installed distribution. 陶芳波’s next milestone is to make the identity-bridge model a de facto standard and improve “symbiosis” and “connection” enough to drive 10x or 100x growth—still a non-consensus, long-duration bet with an unproven commercial loop.

🔗 Original source & video: 125: Using AI to Recreate a “Me”: A Conversation With Mindverse’s 陶芳波 on Identity Models

The 2025 Agent acceleration came from simultaneous advances in code models, RFT, and MCP, with funding and valuations subsequently sending a market signal. Sonnet 3.5 delivered a step change in code generation; OpenAI launched Operator and Deep Research in January and February, respectively; Manus broke out in March, then secured $75M led by Benchmark in May at a $500M valuation. On the day of recording, foreign media also reported that OpenAI was considering a $3B acquisition of Windsurf, while Cursor parent Anysphere raised $900M at a $9B valuation. The deeper meaning of “the real Agent era is here” is that foundational components are beginning to work together—not that a few more chat windows have appeared.

This paradigm shift is not about models becoming more articulate; it is about them learning to adjust their action strategies based on environmental rewards. The Agents of 2023 and 2024 relied on knowledge bases, tools, and human-designed workflows, while the model’s training objective remained conversation. RFT places reasoning models inside environments such as programming and computer operation, allowing them to explore paths, fail, and retry. Tao Fangbo called it “the AlphaGo moment for language models”: reasoning can now be learned from environmental feedback, creating the “big gap” between old and new Agents.

Coding is becoming the universal action layer of the digital world, giving Cursor, Windsurf, and Devin potentially far more upside than traditional IDE plugins. Windsurf’s context engine can recognize code, tests, configuration, the command line, and technical documentation; Devin goes further by bringing the browser, IDE, testing area, and editable “strategy room” and “war room” into the environment. With MCP connections, Cursor could eventually shift from “producing code” to “completing tasks.”

General-purpose Agents could absorb many consumer scenarios that once supported standalone startups, making the defensible opportunities increasingly small and specialized. A single Deep Research could cover academic research and market-research tools; a single Operator could cover grocery shopping, ticket booking, price comparison, and travel planning. The product boundaries controlled by large companies will also be broader than in the App era. Tao Fangbo therefore questions how many service Agents the consumer market can still support, while favoring vertical products in insurance and bidding that have proprietary “weapons arsenals,” evaluation standards, and data environments, as well as “second selves” that express individual identity.

Manus’s advantages begin with productization and controllability; its weaknesses are its general-purpose positioning, data barriers, and broken network effects. Kolento prefers it for broad but shallow research when entering a new field: the process is transparent, users can jump to live and take over at any time, and reports can become interactive websites. Its instruction-based memory is also more behavior-aware than ordinary RAG memory. But “world’s first general AI agent” can attract early adopters while also implying that “generality means having no first-association use case,” and login, permissions, Xiaohongshu, or Facebook data walls are far beyond what even the coding-, GUI-, and command-line “octopus” can penetrate.

The real capability center of an Agent product may not be the prompt but the evaluation layer spanning intent, tool calls, execution, and output. Kolento’s analogy was: “The prompt is the weapon, but evaluation is your crosshair.” Only when every change can be quantified can a Demo become a product that improves continuously. If evaluation against real environments can be upgraded into reproducible rewards, Agents could explore and align themselves; but the discussion noted that RFT may deliver only about a 25% improvement over SFT while multiplying costs, so commercialization must account for the marginal return.

The industry’s tempo will diverge sharply: products are being built “almost frighteningly fast,” while mass adoption and ecosystem formation may still take years. Kolento has shifted from heavyweight PRDs and planning toward faster execution, but also revised “having built more than 200 Agents” to say most were merely chatbots or workflows; Tao Fangbo likewise moved from calling 2025 “the year of agent” to taking a more cautious view. MCP currently looks more like “an AI wrapper around APIs”; certification, privacy, data access, and platform monetization have no unified answers, and opening a technical channel does not mean an Agent can freely enter an environment.

🔗 Original source & video: E191 | Small, Beautiful Opportunities Are Here: On the New Paradigm in This Wave of AI Agent Evolution