Onchain Analyst || Researcher || Trader || Content Creator
Joined July 2023
- Tweets21.6K
- Following1K
- Followers1.6K
- Likes22.2K
Pinned Tweet
after reviewing 97 wave 3 submissions, these are my personal favorite demos.
this is not a ranking, just the projects that impressed me the most because of their technical quality, execution, and attention to detail.
every team that shipped deserves respect, building is never easy.
1. BasisDesk – basisdesk.vercel.app
A delta neutral basis trading platform that earns funding rates while reducing price risk.
- what impressed me: every financial calculation uses fixed point arithmetic instead of floating point, making it one of the most accurate implementations I reviewed.
- best for: traders looking for funding-rate opportunities with lower market risk.
2. SoNarr – so-narr.vercel.app
turns crypto narratives into real onchain trades using SoSoValue data and SoDEX.
- what impressed me: It showed three confirmed mainnet orders with careful checks before execution.
- best for: analysts and index creators who want to automate narrative-based investing.
3. Helix – helix-alpha-kappa.vercel.app
A news-to-signal platform that analyzes headlines and generates trading signals.
- what impressed me: It includes a correlation heatmap to prove which signals actually worked instead of just making claims.
- best for: traders who want transparent and evidence-based signals.
4. Conviction Matrix – conviction-matrix.vercel.app
scores crypto sectors using ETF flows, VC activity, and treasury data.
- what impressed me: uses proper statistical methods like Wilson confidence intervals and verifies predictions over time.
- best for: anyone who values measurable and data-driven signals.
5. Sonar – sonar.my.id
An AI hedge fund assistant that turns ETF flow data into trading ideas.
- what impressed me: the builder openly fixed 34 issues found during an audit instead of hiding them.
- best for: small funds and solo traders who value transparency.
6. SoSoMind – sosomind.vercel.app
A complete AI trading terminal with research, execution, and portfolio tools.
- what impressed me: It never shows fake data. If information isn't available, it simply says "unavailable."
- best for: active onchain traders who want everything in one place.
7. MARA – mara-neon.vercel.app
An ai macro trading agent that records its reasoning onchain.
- what impressed me: strong historical validation with a very rigorous performance evaluation.
- best for: traders who want transparent and verifiable AI decisions.
8. SoSo Analyst – soso-analyst.vercel.app
A protocol that lets analysts build an onchain reputation through verified predictions.
- what impressed me: It showed one of the strongest end-to-end execution examples I reviewed.
- best for: researchers and users who want to follow analysts with proven records.
9. Mosaic – mosaic-ivory.vercel.app
builds crypto index portfolios from simple natural-language investment ideas.
- what impressed me: the team found a major execution bug, fixed it, and added tests to prevent it from happening again.
- best for: Investors who want easy thematic investing.
10. Prism – indexforge-kappa.vercel.app
A quantitative index builder using a deterministic five-factor model.
-what impressed me: AI is only used to understand the investment idea. Portfolio construction stays fully deterministic and reproducible.
- best for: Index creators and quantitative researchers.
These were my personal favorites after reviewing 97 submissions on @SoSoValueCrypto
there were many great projects that didn't make this list, and every builder deserves credit for shipping something.
I'm excited to see which of these projects continue to grow after the buildathon.
which demos impressed you the most? 👇
A few products have really stood out to me so far, not because I'm ranking them, but because each one did something that made me stop and think.
Edgework has been one of the most consistent products across all three waves. Instead of relying on flashy AI claims, it focuses on real data, bootstrap statistics, smart-money positioning, and confidence backed by evidence. trade check feels like a tool that actually helps you make better decisions.
HATCH was probably my biggest surprise, coming into the competition in wave 3 and shipping something this complete isn't easy. the parent-child investing model, clear ownership structure, and smooth user flow made it feel like a product built for real families, not just for a demo.
HEIRLOCK made me pause more than once while reviewing it. the amount of work behind it is hard to ignore, persistent investment memory, AI agents debating before making decisions, learning from every trade, production skills, MCP tools, and onchain continuity contracts. It feels more like something a startup has been building for months than a typical hackathon project.
Prism stood out for a different reason. I really liked the thinking behind its design. Instead of letting AI decide portfolio weights, it lets a quantitative model do the calculations while AI focuses on understanding and explaining the investment idea. It's a simple choice, but a very smart one.
I have also enjoyed reviewing products like Bellwether, SoVibe, BasisDesk, POD, Conviction Matrix, ETFPulse, SOSO DNA, SoDEX Tracker, SoSoFlows, and SoSoMon. They all took different approaches, but the common theme was that they tried to build something people could actually use and verify, instead of just talking about what they planned to build.
I'm still working through the remaining wave 3 submissions, so I'm sure there are more products that will surprise me before I'm done.
thanks for the opportunity 🙏
@SoSoValueCrypto @BTCtensai
ONE win wasn't enough for N3on
Bro went straight back for another on Jack.com like it's nothing @N3onreload
Bro is so drained he barely even reacts to a $7,000 WIN, N3on on Jack.com is a whole different vibe @N3onreload
DAVIS (❖,❖) retweeted
A supplier can give you the RIGHT answer in three different languages and still lose a deal because a blockchain sees three different answers.
That is the problem @GenLayer is tackling with its EQUIVALENCE PRINCIPLE.
I learned something similar from solving problems: the rule that makes a system reliable can also limit what it can handle.
Bitcoin solved double-spending by making independent nodes agree on one ordered history. Alice cannot spend the same coin with Bob and Charlie if the network agrees on which transaction happened.
Every node must reproduce the same result. That precision is essential for payments, but it becomes a problem when a task needs judgment.
Two AI models can answer the same question differently without either being wrong. Traditional blockchains cannot simply treat those answers as interchangeable.
GenLayer changes what validators check.
The video puts it simply: machines used to agree on numbers. Now, they have to agree on meaning.
Through the Equivalence Principle, an Intelligent Contract defines which differences are acceptable.
- An exact price can require an exact match.
- A measurement can allow a specified tolerance.
- A delivery dispute can accept different wording if the substantive conclusion is the same.
A leader proposes an answer. Other validators check it independently against the contract's rules. They do not need to produce identical sentences. They need to determine whether the proposed result meets the agreed criteria.
If they disagree, the protocol has mechanisms to handle that disagreement instead of blindly accepting the leader's answer.
The flexibility comes from the contract, not from letting AI decide its own rules.
Now imagine your business serves clients who speak different languages.
One asks in English. Another asks in French. A third asks in Yoruba.
They all want to know the same thing: How many tons of cocoa can you supply per week?
The wording changes. The commercial question does not.
I would explore GenLayer's Equivalence Principle here first: checking whether different responses communicate the same agreed quantity while keeping the actual supply figure within the contract's rules.
Sometimes, the challenge is not getting the right answer. It is building a system that recognises it when the answer sounds different.
Where would you deploy this first: multilingual business agreements, delivery disputes, or another real-world problem?
DAVIS (❖,❖) retweeted
A builder in a group chat had the token live and a list of accounts that never answered. The product was finished. The explanation was still sitting in his notes.
That is the distribution problem. Attention in crypto usually waits on a yes from someone you do not know. Permissionless Campaigns are now live on @RallyOnChain, and anyone can open one in about five minutes, funded with stables or the project’s own token. You write the brief, fund it, the money sits in escrow, and the campaign goes live. Anyone can open one. That is the part that was missing.
Write the brief before you fund it. A campaign with no sentence is just a balance.
If the campaign can be paid in the project’s own token, is the hard part still the budget, or getting a stranger to explain the thing?
Watch the full livestream xitter.cf/i/broadcasts/1pJkOXQNj…
Watch the full livestream xitter.cf/i/broadcasts/1pJkOXQNj…
Watch the full livestream xitter.cf/i/broadcasts/1pJkOXQNj…
Watch the full livestream xitter.cf/i/broadcasts/1pJkOXQNj…
Watch the full livestream xitter.cf/i/broadcasts/1pJkOXQNj…
The interesting part isn't AI writing another resume. It's your agent opening the employer's site, filling the application, answering questions, and attaching the tailored resume. @dreamworkhq
Introducing the Dreamwork MCP 🤖
Your favorite agent can now get you your dream job.
It gives you access to 2M live roles, perfectly tailored resumes, and most importantly: automated submissions directly to the employer.
Paste this: npx -y @dreamworkhq/mcp
So your agent can now search live roles, write a tailored resume, answer screening questions, and submit the application on the employer's own site. @dreamworkhq
Introducing the Dreamwork MCP 🤖
Your favorite agent can now get you your dream job.
It gives you access to 2M live roles, perfectly tailored resumes, and most importantly: automated submissions directly to the employer.
Paste this: npx -y @dreamworkhq/mcp
Watch the full livestream xitter.cf/i/broadcasts/1pJkOXQNj…
Watch the full livestream xitter.cf/i/broadcasts/1pJkOXQNj…
Watch the full Sui Basecamp livestream: xitter.cf/i/broadcasts/1pJkOXQNj…