The rabbit hole is deep with these models. The more you dig, the deeper the hole gets, and the more you realise that the Dunning-Kruger effect is certainly strong with most YouTube trade "expert videos". So much buillshit! The delay has been my battle to fix issues with a paper trading platform I built a while ago. I've learnt that AI unintentionally lies. It wants to reach an answer and, when stuck, fabricates outcomes. Adversarial code reviews and cross-model evaluations are highly recommended. That said, in this volatile bull-bear market, there is certainly money to be made, and in my mind, as expertly validated by my cohort of friendly assistants (four dogs, a laptop and a few AI models). Hopefully, soon the trading balance will begin to reflect the approach. For now, I'd better get to day 17 before I get sucked into the black hole of bug fixing something I now feel can't be repaired. That said, Gemini, Claude, and even Grok told me it worked until they started reviewing each other's work. Peace out, Agentic adventurers.

Finally up to day 15, but I keep getting pulled into the vortex of trying to resolve old issues. Love these knowledge bases and that Claude pulls me up on contradictions.

From time to time, I'll get an idea that changes how I run my business. But having talked with the Founding 12 over the last 2-4 weeks, I've got something that will change your business/income too. @Derek Charlier , a Founding 12 Member of The Accelerator Program suggested that any student taking part in the Agents to Income course wants a "Quick Win". Something they can take that proves beyond a shadow of a doubt that they CAN make money with ai. Well I've got 2 amazing candidates for this. Without spoiling anything, I have created two businesses that will run for you automatically, with a single prompt that have the potential of making $250k+ per year. Oh it's so cool. If you want to hear about it before I launch, comment "AWESOME" on this post.

A week or two ago, I shared that I had turned my localhost crypto scanner into a live web beta after several members asked if I had a URL. Since then, I’ve made a pretty major update. CoinPulse is no longer just scanning Robinhood-tradable cryptos. It now cross-checks crypto movement across multiple data sources/exchanges, including Binance, Coinbase, Kraken, and CoinGecko, so the system can better tell whether a move is broad-based or isolated to one venue. I also redesigned the landing page and dashboard around a clearer question: Is this move confirmed, liquid, and worth watching? New additions include: - Multi-exchange confirmation - Liquidity and spread warnings - Candidate Pipeline page - Market Intelligence page - System Performance / self-grading analytics - Blocked-signal tracking - Missed-profit tracking - Custom trading-style presets - Watchlist settings - Cleaner Simple / Pro dashboard experience - Pricing and account settings pages The part I’m most excited about is that CoinPulse is starting to behave less like a simple crypto scanner and more like a research assistant. It does not just say “buy” or “watch.” It explains why something is trade-qualified, still in the pipeline, blocked, or not worth chasing. This entire project is still being built with Claude, ChatGPT, and other AI tools.... and continues to be heavily inspired by Lewis’ videos and the daily build updates from members in this community. Beta is live here: https://getcoinpulse.com Would love feedback on the new positioning, dashboard flow, and whether the multi-exchange confirmation makes the product feel more useful.

Building a world-class AI that resonates with what I like, and how I like it. This is so exciting