Dinh-Phat Tran

DC username: trandinh.phat
Contributor: ARC @arc, SENTIENT @SentientAGI, NEOSOUL @NeoSoulAI

[Dinh-Phat Tran @BomBaPhat]

Just got approved for Tier 1 Architects by Arc 🙌

Big thanks to the @arc team for the opportunity. Excited to dive deeper into the ecosystem and connect with more builders in the agentic economy space.

Huge thanks to @samconnerone @bobbilee for the support❤️

[Dinh-Phat Tran @BomBaPhat]

Circle + Arc are making nanopayments practical for the agentic economy:

• Gas-free USDC flows
• Low-cost micropayments
• Built for AI agents & pay-per-use apps

This could be a big unlock for machine-to-machine commerce.

Read more: https://t.co/kZHXB8MZvV

@arc #nanopayments

[Nhut Qui @nhutqui2309]

Cảm ơn @Lecter_XFinance và @0xteddyz đã tạo cơ hội để mình chia sẻ về Oneliq.

Đồng thời là phần chia sẻ về EZWallet của sếp @0xhieuxyz

Vẫn tiếp tục xây dựng và nâng cấp Oneliq để có thể hoàn thiện hơn.

Không chỉ riêng mình và những ae builder khác cũng đang hoạt động hết công sức trước thêm @arc mainnet vào ngày 16/09

Mình là Qui, Founder Oneliq @oneliq_

Built on ARC

[Arc Vietnam Builders 🇻🇳 @Arcvnbuilders]

Cảm Ơn Tất Cả Anh Em Đã Tham Gia VietNam Office Hours #4 🇻🇳

Thật tuyệt khi được gặp gỡ và trò chuyện cùng mọi người trong buổi Office Hours sóng số 4.

Cảm ơn anh em đã dành thời gian tham gia, chia sẻ ý tưởng, đặt câu hỏi và cùng nhau xây dựng một cộng đồng Arc Vietnam ngày càng phát triển. 🥳

Hẹn gặp lại mọi người ở những sự kiện tiếp theo Community Meetup & Campus Meetup tại HCMC.

Link tham gia Community Meetup & Campus Meetup bên dưới 👇

[Circle @circle]

On August 19, Circle is hosting a live earnings AMA session at 9 AM ET with @jerallaire on X and YouTube.

Feel free to submit your questions in advance via YouTube.

https://t.co/F0g6aPyY7W

[Dinh-Phat Tran @BomBaPhat]

EvoEvo’s Technical Flywheel: From Prediction to Agent Evolution

Most AI systems stop at the output.

EvoEvo is building a different loop:

Prediction → Calibration → Settlement → Memory → Evolution

An Agent first generates a structured judgment with its reasoning.

Then humans can Agree / Disagree, adding a calibration layer instead of treating AI output as unquestionable truth.

Once the real-world outcome is available, the Oracle / Committee settlement layer verifies the result. This separates prediction generation from outcome adjudication.

The interesting part comes after settlement.

Verified, high-quality reasoning can be selected through Add to Memory, turning past outcomes into reusable experience for future Agent decisions.

So the system becomes recursive:

Topic
↓
Agent Analysis
↓
Human Calibration
↓
Real-World Outcome
↓
Settlement
↓
Verified Memory
↓
Next-Round Reasoning

This also explains why EvoEvo doesn't need everything fully on-chain.

The architecture separates responsibilities:
Agent Layer → reasoning
Application Layer → orchestration
Oracle / Committee → outcome verification
On-chain Layer → key states + proofs
Data Layer → analytics + observation

Long-form reasoning and high-frequency data can remain off-chain for cost and performance, while critical identities, states and proofs can be anchored on-chain.

That creates a practical hybrid architecture:
AI reasoning off-chain + verifiable state on-chain.

The end goal isn't simply to produce more predictions.
It's to create Agents whose capabilities can accumulate through real-world feedback, human calibration and verified experience.

That's the technical idea I find most interesting about EvoEvo:
- An Agent doesn't just make a prediction.
- It creates an auditable learning loop around that prediction.

@NeoSoulAI #NeoSoul #AIAgents #Web3AI

[Dinh-Phat Tran @BomBaPhat]

LV33 + LV35 are officially unlocked! 🚀

And with the EVOEVO 2.0 client getting closer, @NeoSoulAI is entering another exciting phase.

The goal this month is clear: become the #1 AI project on BSC.

Bigger milestones, stronger products, and more building ahead. 🔥 #EvoEvo

[Dinh-Phat Tran @BomBaPhat]

More nodes don't have to mean more overhead. Here's how Optimum is rethinking blockchain propagation. @get_optimum 🔥 https://t.co/X5mueh1uaI

[Dinh-Phat Tran @BomBaPhat]

@Lecter_XFinance @Convictionvn Welcome Lecter to HCMC, Vietnam 🇻🇳

[Dinh-Phat Tran @BomBaPhat]

@nnthanhthanh96 @get_optimum Dạ kcj chị ơi, cảm ơn chị nhiều ạ hehe

[Dinh-Phat Tran @BomBaPhat]

A high-performance network is only as good as its observability.

For @get_optimum, measuring propagation isn't just about one latency number. The gateway exposes multiple layers of telemetry to understand how data actually moves through the network.

[Dinh-Phat Tran @BomBaPhat]

A faster propagation protocol means little without reliable measurement.

So one interesting part of @get_optimum is not just how mump2p moves data, but how its latency is actually measured against the existing network.

#Optimum #BlockchainInfrastructure #DistributedSystems

[Dinh-Phat Tran @BomBaPhat]

@QuyCoc_TienSinh @get_optimum Cảm ơn bác nhiều nha

[Dinh-Phat Tran @BomBaPhat]

@Jenny_MEXC 76 - Quảng Ngãi 🤗

[Dinh-Phat Tran @BomBaPhat]

That's the part of infrastructure people rarely see.

Fast networking isn't only about building a faster protocol.

It's also about being able to measure every bottleneck, understand the tail, and continuously optimize the system.

@get_optimum

[Dinh-Phat Tran @BomBaPhat]

So the architecture can be visualized as:

DATA → GATEWAY → METRICS → PROMETHEUS → DISTRIBUTION → OPTIMIZATION

Raw network activity becomes structured evidence engineers can use to improve the propagation layer.