What progress have you made since your last update?

本週我們新增了以下幾個功能：

1.AI統一帳本：該功能不是只有面對投資者而是全方面的，除了也可以用作傳統方式開表單日常記帳外，也可以搭配相關AI工具進行財務紀錄近日的收入支出等紀錄，同時配合 Human in the loop ，需要人為介入進行相關記錄新增/編輯/刪除等操作，而這概念也是目前好比於公司訂單、線上消費...等涉及到金流或者利益都會需要的重要功能

2.記憶模組優化：記憶功能是我們長期以來優化的目標，不只追尋個大開源加以整合，也在審視如何整合的效果會是會好，此次我們新增有時效性管理，以及搭配來源追朔與衝突處理等可以監測”最新記憶“或者“已被取代的記憶”來讓使用者更好從龐大的記憶中區分自己觀眾的記憶點。

3.Skill版本紀錄：如同記憶模組一樣重要，Skill也新增了支援完整版本軌跡，來增加使用者的識別度。

4.新增訪客模式：為了能夠讓沒有TON或者區塊鏈錢包的相關使用者能夠也有機會體驗我們平台，新增了訪客模式，但由於眾多功能都還是需要AI協作，但我們能夠承受成本有限的情況下，目前流量我們也會嚴格控管。

在此，希望看到我們所做，並且肯定我們所做的人，可以給予贊助，我們也願意將資金用平台維護/開發，以及更廣泛的免費推廣，以及邁向有節制的落地模型免費提供用於平台內部，以及資料本地部署整個平台服務。

Google Translate:

This week we've added the following features:

1. AI Unified Ledger: This feature isn't just for investors; it's comprehensive. Besides traditional forms for daily accounting, it can also be used with AI tools to record recent income and expenses. Human-in-the-loop functionality allows for manual intervention in adding, editing, and deleting records—a crucial feature for transactions involving cash flow and profits, such as company orders and online spending.

2. Memory Module Optimization: Memory functionality has been a long-term optimization goal. We've not only integrated with major open-source libraries but also examined the best integration methods. This time, we've added time-sensitive management and features like source tracing and conflict resolution to monitor "newest memories" or "replaced memories," allowing users to better distinguish their audience's memories from a vast database.

3. Skill Version Recording: Just as important as the memory module, Skill now supports complete version tracking to increase user recognition.

4. New Guest Mode: To allow users without TON or blockchain wallets to experience our platform, a guest mode has been added. However, since many functions still require AI collaboration, and given our limited budget, we will strictly control traffic.

We hope that those who see and appreciate our work will provide sponsorship. We are willing to use the funds for platform maintenance/development, wider free promotion, and a controlled implementation model, providing these resources free of charge for internal platform use and local data deployment.

What are your next steps?

1. 我們希望能夠提供更全面的理解自己的支出/收入等情形

2. 新增日曆系統可以讓使用者對於未來的資金進行規劃，並且標示，同時日期區進行可以告知使用者

3. 新增AI對於自己的相關日常開銷/收入，給予相關性的建議

如果我們有足夠資金贊助可能會進行以下開發：

1. 語音模式輸入支持

2. 更多跨平台支持（目前僅telegram)

3. 提供落地模型服務租借

4. 提供個人檔案/知識庫管理

Google Translate:

1. We hope to provide a more comprehensive understanding of your spending and income.

2. The newly added calendar system allows users to plan and mark future funds, while date zones provide information.

3. The added AI provides relevant suggestions regarding your daily expenses and income.

If we have sufficient funding, we may undertake the following developments:

1. Voice input support

2. More cross-platform support (currently only Telegram)

3. Provide on-premises model rental services

4. Provide personal profile/knowledge base management for AI

What progress have you made since your last update?

1.添增募資板塊外部分享連接，可以讓使用者看到每次募款者，可能針對投資人等問題後進行修改的軌跡。

Adding external sharing links to the fundraising section allows users to see the trajectory of each fundraiser's actions, including potential modifications made in response to investor inquiries.

https://cryptomind-ton.zeabur.app/static/studio-share.html?t=sh_BWj1dDAcJI_g3O4qxJ6CY9aV

2.增加AI審核功能，可以替使用者評估該專案募資風險，是否值得投資，以及可以配合產生後的否險報告宜動到聊天室後更近一步的相關詢問。

The addition of AI review functionality can help users assess the fundraising risks of a project, determine whether it is worth investing in, and allow users to follow up with further inquiries in the chat room after a risk report is generated.

What are your next steps?

1.當有人使用後我們想建立一個管道，可以搜集相關專案的使用者與AI討論後的結果，使用者不一定要分享，但分享後可以看到其他人對於該專案的評價，或者困惑點，提供給其他捐款人進行參考

Once someone uses it, we want to establish a channel to collect the results of discussions between users and AI regarding related projects. Users don't necessarily have to share, but if they do, they can see other people's evaluations of the project or points of confusion, providing other donors with reference.

What progress have you made since your last update?

這個週末我與夥伴在探討者，目前AI創作文章眾多，但我們真的理解我們的需求是什麼的，有時我們會有想法但卻沒有目標，這往往是最困擾我們的事情，在這個AI盛起的時代，很多人會選擇直接丟給AI來產生一套完整論述，但無法評判這個過程，是否是符合自己當初所設想的。而這也會給審批人造成勞力上，甚至思想上的疲勞。

我們不排斥與AI討論，但如何正確地AI與自己討論，又該如何使審批人能夠從這茫茫大海中的文獻中，知道這篇文章，或者創意是可行的、可信的，我們認為思考中的轉述是很重要的，如果對於某項領域是真的有自己理解，制定自己解決方案中的轉折確實會有，但這個過程式層遞式進行，不太可能因為一個轉念就大改自己的專案計劃，或者自己所想，而這也是完全使用AI工具的人常會犯的錯誤，因為只在乎頭乎，卻忽略了每次過程中提出的最佳解，只是一次又一次的用文章覆蓋上一次的層果。

所以我們做了個假想實驗，我們保留AI與人討論的範圍，但是AI主要會針對我們所給出的議題提出反問，類似於TypeScript 教學大神 Matt Pocock 所寫的 grill_me，但我會記錄下每次使用者詢問AI的差異性，類似於Github diff，而我們也認為這個將可以賦予審批者一個蠻重要的參考訊息，提案者思想過程中的差異性，讓審批者能夠理解使用者的不管是起承轉合是否合理，又或者只是一鍵生成貼上的AI內文。

Google translation:

This weekend, my partner and I discussed the current abundance of AI-generated articles. However, we don't truly understand our needs. Sometimes we have ideas but lack a clear objective, which is often our biggest frustration. In this era of AI proliferation, many people choose to simply dump their ideas onto AI to generate a complete discourse, but they can't judge whether the process aligns with their initial vision. This also leads to mental and physical fatigue for reviewers.

We don't reject discussions with AI, but how can we ensure that AI engages with us effectively? How can we ensure that reviewers can glean from the vast sea of ​​literature whether an article or idea is feasible and credible? We believe that paraphrasing during the thought process is crucial. If one truly understands a field, there will be turning points in developing one's own solutions, but this process is hierarchical. It's unlikely that a single shift in perspective will drastically alter one's project plan or original ideas. This is a common mistake made by those who rely solely on AI tools—focusing only on the initial idea while neglecting the optimal solutions developed at each stage, simply covering up previous results with articles.

So we conducted a hypothetical experiment. We maintained the scope of the discussion between AI and humans, but the AI ​​would mainly ask questions about the topics we provided, similar to grill_me written by Matt Pocock, a TypeScript tutorial guru. However, I would record the differences in each user's questions to the AI, similar to Github diff. We also believe that this can provide reviewers with important reference information: the differences in the proposer's thought process, allowing reviewers to understand whether the user's ideas are logical or simply a one-click generated AI text.

What are your next steps?

• 目前功能皆已經開放，但因為平台性，還是架設在原TON進行開放，我們也希望將平台從原本的單純金融平台，擴展一個更有意義的，為人類/AI協調提供更好的協作方法與特性。但首先我們還是希望有看到這則貼文的人，可以嘗試使用我們的平台，或者資助我們，提供我們後續開發援助。也許我們不是世界上最頂尖的一群工程師，但我們願意且努力地想尋找我們能夠發揮的舞台，因為我們都知道AI絕不只能是僅限於當下自己所知道的東西。

Google translation:

All functions are currently available, but due to the platform's nature, it's still hosted on the original TON platform. We hope to expand the platform beyond a simple financial platform to provide better collaborative methods and features for human/AI coordination. However, our first priority is that anyone who sees this post will try using our platform or sponsor us to provide further development assistance. We may not be the world's top engineers, but we are willing and working hard to find a stage where we can contribute, because we know that AI is far more than just what we currently know.

Is there anything others could help you with?

• 來自於Austin Chen討論的以及與夥伴後續討論誕生出得想法。

Google translation:

These ideas originated from discussions with Austin Chen and subsequent discussions with partners.

What progress have you made since your last update?

Spent the last two days fixing the skill/memory feature — turns out it was completely broken after it went live. Users could ask the agent to create or manage skills in chat, but nothing actually happened. The agent just wrote text back saying "done!" when it hadn't done anything.

Dug into it and found 6 separate bugs stacked on top of each other. Each one alone would've broken the feature, but together it was totally dead:

• The agent literally didn't know the tools existed (no prompt guidance)

• Even if it did, the tools weren't registered in the DB so they'd get filtered out anyway

• The consent gate that's supposed to pop up a confirmation card? Its trigger logic was unreliable

• Asking "what skills do I have?" would accidentally trigger an investment clarification popup (super confusing UX)

• And a bunch of other small stuff

Also built memory deletion, which was supposed to be there from the start but got missed. So now users can both save and remove things the agent remembers about them.

Fixed everything and tested on production with real conversations — create, list, delete skills, save and forget memories, all going through the consent flow properly. Tested with both deepseek and nemotron models to make sure it's not model-specific.

What are your next steps?

• Keep an eye on whether people actually use the skill/memory stuff now that it works

• The prompt is pretty strict but weak models (like deepseek free tier) still occasionally pretend they did something without actually calling the tool. Might add a UI hint suggesting stronger models for this feature

• There's a missing DB table for audit logging that I keep running into — not urgent but should fix eventually

Is there anything others could help you with?

• The deploy platform (Zeabur) sometimes can't reach GitHub during builds — nothing to do with our code, just network hiccups. Usually fixes itself on retry

• I've got an approximation for user retention stats but doing it properly would need a new table. Not blocking anything, just a nice-to-have for later

Project Update v1.0.1 — Changes from v1.0.0

I want to share a short update about what has changed from the original proposal.

1. Production swap execution

CryptoMind production will stop preparing executable swap transactions. Even though users sign with their own wallet, the platform still had an important role by getting the quote and preparing the transaction message. I decided to reduce this responsibility instead of adding more production swap pairs.

The complete risk-gated swap flow, including amount limits, consent records, replay protection and the Safety Kernel, will remain available in testnet for research, security testing and independent review.

2. Product focus

The production platform will focus more on crypto and stock analysis, news comparison, scam detection, read-only wallet risk checking and community reporting.

Optional AI subscriptions may later support more queries and advanced analysis. Our team believes AI itself is not the main problem. The more important question is what tools, permissions and execution ability we give to the AI.

3. Funding use

Funding that was originally planned for expanding production swap pairs will instead support:

- independent review of the Safety Kernel and testnet workflow;

- a public Agentic Money Safety benchmark;

- adversarial testing;

- TEE deployment and attestation; and

- public documentation and reusable safety interfaces.

If funding exceeds the original target, the main goals above will still come first. After that, I may compare rented infrastructure with dedicated hardware for hosting selected open-source small language models, increasing AI query capacity and supporting users outside the TON ecosystem.

4. Early supporters

After the membership service becomes stable, I hope to provide eligible early supporters with a non-transferable Founding Supporter membership. The exact features, reasonable usage limits and eligibility will be published before it is provided.

I am sharing this update before the grant approval process is completed, so supporters can review the current scope and decide based on the updated direction.

Thank you for the feedback and early support. I will also update the main proposal and share the production result after the conservative-mode update is deployed.

By the way, just to add one quick thing I forgot to mention: besides TON, we're actually also looking to expand into Pi Network and other ecosystems. But since everyone on the team has their own main job to focus on right now, building out these ecosystems is something we'll plan for the future when we have more time and capacity.

Austin Chen Thank you so much. I really appreciate this.

Honestly, the amount is not the important part to me. Even if it were just one dollar, knowing that someone sees what I am trying to build and thinks the idea is worth supporting would already mean a lot to me.

CryptoMind has taken a lot of my time over the past 8 months, and most of the time I am just building and testing things by myself. So when someone understands the idea behind “agents with money, safely” and is willing to support me personally, I am genuinely very touched by that.

I also appreciate your honesty about not being especially interested in TON or crypto. For me, TON is mainly where I am testing this idea right now. What I really want to explore is whether AI agents can interact with real money in a way that does not simply depend on the LLM making the right decision.

Thank you for recognizing what I am trying to do. I will remember this support.

One more thing I want to mention is that we are not a famous team. Actually, I am not even sure if I should call us a formal team yet.

CryptoMind started as a project I built by myself, and later some friends joined to help with different parts. We don't have a company behind us, large funding, or many resources. Most of the development until now was done with our own time and money.

But we already spent a lot of time trying to make the project actually work, not only making a concept or demo. What we hope now is to have a chance to continue pushing it further and see how far this idea can really go.

Funding would help us with things that are difficult for a small project to do alone, such as security audit, better infrastructure, TEE deployment, testing and running a larger community.

Support also does not necessarily need to only be money. Server credits, computing resources, hardware, infrastructure support, security support, or opportunities to test and work with other teams would also be very useful for us.

We know CryptoMind is still small and there are many things we need to improve. But we hope to get a chance to actually try these ideas in the real world, learn from what works and what does not, and keep developing the project instead of stopping at the current stage.

If people or organizations think this direction is worth trying, any kind of support that helps us continue building and testing it would mean a lot to us.

I also want to share two things that are not the main part of the proposal, but are related to what I want to do next.

1. If there is more funding, I want to open the community features

In the proposal, I mainly put the funding on security audit, testnet testing and TEE deployment, because these are the things I can do immediately and also easier to verify.

If the project can reach the funding goal and still have enough resource, I also want to upgrade the server and open some community features that are already in CryptoMind.

These include forum, friend system and governance features.

Actually these functions are not new ideas. I already built most of them around February and March 2026, but I did not open them to public users.

The reason is quite simple. I don't think it is a good idea to open a crypto forum before I have enough identity and anti-scam protection. If anyone can easily create many accounts, the forum can become full of spam, fake promotion or scam very quickly.

So I decided to keep these features hidden first instead of opening them just because they are finished.

2. Longer term, I want to make the trust system more useful for users

One problem I have been thinking about is why users should continue using CryptoMind.

AI agent itself is probably not enough.

Now many projects can build agent, tool calling and skills. The frameworks are also becoming easier to use, so I don't think "we have an AI agent" can become a long-term advantage.

CryptoMind already has some trust and risk-control logic in the backend. My next direction is to make part of this system visible and useful to users.

One idea is proof of personhood.

I am considering Human Passport, previously Gitcoin Passport, as the first integration for reducing sybil accounts. Later I may also support World ID as another verification option for users who want stronger verification.

Currently the trust score in CryptoMind is still quite basic. In the future, I want the score to include more real signals.

For example, users may have different trust levels or badges. Other people don't need to see all private details, only a simple trust level. But the user can see their own score details and understand why the score goes up or down.

I also don't want the score to only increase forever. If the account is inactive for a long time, the score can slowly decrease. If a connected wallet is reported or flagged by risk systems, it can also affect the trust score.

Another idea I am exploring is reputation-based co-signing.

For example, a user with higher trust level may be allowed to help co-sign or approve some transactions. This can give the reputation system an actual function instead of only showing a badge.

I also like this direction because it does not require CryptoMind to create its own token.

The reason I am thinking about these features is related to the user adoption risk I mentioned before.

If I only open forum, friends and governance, I don't think that is enough. There are already many communities that can do the same thing.

But if the community has some way to reduce bots and fake accounts, and users can know they are more likely talking with a real person, I think it can become more useful, especially for financial and crypto discussions.

I don't want CryptoMind forum to become another place full of bot comments, fake signals and scam links.

I am currently writing these ideas into a design document. I am also still deciding which proof-of-personhood solution is the most suitable, and whether reputation-based co-signing is practical enough to build.

Feedback on these two parts would be very helpful.