AI systems I’ve built
What I’ve built, and how each one works.
Vano
A Telegram trading assistant with three products: Hunter for managed trade calls, Quant for market analysis, and an assistant that reads charts and answers from live data.
- 2,900+ users
- +15.13R across 22 closed Hunter calls
- Analysis across 9 market categories
- Delivered inside one Telegram app
gtmquant
An AI go-to-market engine. One product brief goes in, and a team of specialist agents returns one strategy you can execute.
- Positioning, ICP, messaging and channel strategy
- A 90-day execution plan delivered as a report
KeepFit
A personal trainer app for home and gym workouts. Ten questions produce a weekly plan with safe weights, form guides and meals from where you live.
- 53 exercises with male and female form guides
- Meal plans across 45 countries
- No account, all data stays on the phone



ADSWARM
A media buying system of 20 AI agents across the Google, Meta, TikTok and Snapchat ad platforms, running from strategy and creative through to reporting.
- Covers strategy, creative, buying and reporting
- Results feed back into new creative and budgets
SEO Audit Platform
Ten AI agents turn a site URL into a ranked list of fixes.
- Technical, on-page, keyword and backlink audits
- Each fix ranked by impact and effort
Hyperliquid trading bot
An automated perpetuals bot for Hyperliquid that finds trades, manages risk and learns from its own history.
- Alerts and position controls on Telegram
Also built
- memecoins-trading-agentTracks smart wallets, scores tokens and executes via Jupiter.GitHub ↗
- content-creator-skillsThe content skills and agent workflows I use to publish at volume without a team.GitHub ↗
- motion-graphics-promptsCopy-paste prompts that make a polished product video with a coding agent, in 23 versions.GitHub ↗