Clyde

DevRel @trylimitless | prediction markets, agentic engineering 🇭🇰

[Clyde @clydedevv]

New Video: Train AI Agents on High-Volume Twitter Data

• Scrape tweets (no API keys)
• Train an @elizaOS agent on an individual account's history
• Merge styles from your favorite KOL's (I used @dabit3 + @alfaketchum)
• Fine-tune a model with the data using @togethercompute

[Clyde @clydedevv]

@Panterra0x @heysealai Glory to Moolius Aurelius

[Clyde @clydedevv]

Timeline keeps giving me x402 fomo

[panterra ☀️ @Panterra0x]

If you’re in marketing, you’re cooked

@heysealai just built me a full GTM strategy in seconds from a few simple questions:

channels, content, budget, positioning, next steps

Best part: it uses x402 + USDC, so instead of subscriptions for Exa, Parallel, Claude, etc, you just pay per use

No API secrets

This is what agentic software should feel like

Huge thanks to @echulshin for letting me write this skill and @MooliusC for letting me test it on his peptide coffee business

[echul @echulshin]

This is the future of portfolio management. I taught @heysealai to monitor my stocks & crypto 24/7.

My agent generates an overview of my portfolio and compares my performance against the S&P 500.

I'll drop the exact stack I used to build this below so you can copy it.

[Clyde @clydedevv]

It's time

[Limitless @trylimitless]

User-Generated Markets are LIVE on Limitless

We’re fundamentally reshaping how users interact with prediction markets:

1. Pick an asset
2. Choose an outcome
3. Launch a market

Every creator earns a fee share from its trading activity

Create yours → https://t.co/qYwZiXHCcD

[Clyde @clydedevv]

@TimHaldorsson Hey hey would be happy to get connected. Also gonna steal your tweet

[Lew @Lewsiphur]

My very first product.

Proud to announce I am the Co-Founder and CMO of Hunch.

Months of work has been put into creating the best possible trading experience for prediction markets.

I am so proud of my team and what we have built.

Built by traders, for traders.

[Clyde @clydedevv]

Community eats first

[Limitless @trylimitless]

$LMTS Season 3 Airdrop Claim is LIVE

Season 3 was our biggest one ever, and it’s getting our biggest airdrop ever

To match a record-breaking season, we’ve allocated 3% of the total $LMTS supply to reward our traders

Claim here → https://t.co/iA1jixNx4J

[Clyde @clydedevv]

@arespro Actually crazy that you got Donald trump and Joe Biden to make this vid with you guys

[ARES @arespro]

They're already making films about a hantavirus outbreak

[Clyde @clydedevv]

@shill_ivey @trylimitless Pleasure to get connected boss 😎

[shirish @shiri_shh]

First OpenClaw. Now Hermes 😭

Hermes agent (open source, months old):

- reviews past conversations
- builds skills from experience
- persistent cross-session memory
- gets smarter the longer it runs

Claude "dreaming" (shipped today):

- reviews past sessions
- extracts patterns
- persistent memory
- gets smarter over time

next on the watchlist: whatever Hermes or Openclaw ships in Q3.

[CJ (晨杰) @cjhtech]

Hello world 🇺🇸

[Clyde @clydedevv]

@jordan_ross_8F CLAW

[Clyde @clydedevv]

Spring 2026 prediction market research digest
(incl. Oxford, Imperial College London & Daedalus Research)

tl;dr: Prediction market making is maturing fast. AI agents are becoming the new MMs, spreads are tightening, and depth is up ~30%.

🔹 Calibrated Agents as Market Makers (Shah, SSRN)
https://t.co/uIUgPvXaTl
AI works on long-tail events - but only if calibrated. Raw accuracy gets crushed by adverse selection.

🔹Bootstrapping Liquidity in BTC Prediction Markets (Shabashev, arXiv)
https://t.co/A4kTcJhq63
Cross-market making beats AMMs & DeFi redirection. Risk stays with pros, no impermanent loss for users.

🔹 Institutional Liquidity in Prediction Markets (Dalen / Daedalus, arXiv)
https://t.co/nffOpMdseW
Spreads tighten, depth +30%, but retail can lose in shock states.

🔹PredictionMarketBench (Arora & Malpani, arXiv)
https://t.co/Xg25X1GR4C
SWE-bench-style framework for backtesting LLM/algo agents on real Polymarket/Kalshi data.

🔹 (Tsang & Yang, arXiv)
https://t.co/Pu0qvozK2P
2024 election data: agents that adapt fast to shocks win.

[Clyde @clydedevv]

Hong Kong back on the rise

[CJ (晨杰) @cjhtech]

Capital markets in Hong Kong are experiencing a boom driven by AI

A strong network effect is forming. Over 85% of the AI companies in China that went public this year listed in Hong Kong

[Clyde @clydedevv]

@rbthreek @cjhtech @trylimitless

[Clyde @clydedevv]

@Chance_ @trylimitless Good work team lfg

[Clyde @clydedevv]

@Hermes_HK @AIFlow_ML You gonna be here @AIFlow_ML ?

[Clyde @clydedevv]

Hermes Agent + Mem Palace + Claude code sidecar + Openclaw VPS subagents has me feeling the same excitement of pre-nerf Openclaw

[Alex Finn @AlexFinn]

OpenClaw and Hermes agent are INCREDIBLE together

It's not about using one or the other. It's about learning how to use them as an AI agent team

In this video I go over the strengths and weaknesses of each, plus show you 4 workflows that will skyrocket your productivity:

[zaimiri @zaimiri]

Episode 1: Building a trading bot

Apparently everyone is building Polymarket bots and making millions.

(Spoiler: they aren't - it's bait)

Since Polymarket is overcrowded, I decided to build on Limitless.

Let's see what we can do:

Limitless have has a starter repo with 3 strategies:

→ Oracle arb using Pyth price feeds
→ Signal sniper with CoinGecko spot data
→ Cross market arb when YES + NO < $1

The direction I'm going: Oracle Arbitrage.

The signal is objective. The edge is quantifiable. When Pyth says BTC closes above $100K and the Limitless market is pricing 60%, there's a real number to work with.

First thing I noticed: prediction markets tend to overshoot.

Everyone guesses the same direction.

That's our edge.

There's a lot of opportunity in those momentum based bets.

Not planning to get rich in a few bets but consistent wins over time.

Let's see how this goes.

Planning to have the bot live by next week on @trylimitless 👀