MCG Live · March 16, 2026 · 28 min
GSD 2.0 is a fully autonomous AI coding agent built on the open-source Piper framework that can run for 24 hours unattended, now with Stripe integration for monetization and a UI for non-technical users. Founder Lex explains how the tool's power comes from its opinionated system prompts and custom tools rather than just the underlying model, and plans to monetize through a subscription UI while keeping the GitHub repo open-source—GSD already has 28k stars and is seeing rapid adoption.
Bags 100k challenge: 50k to developers and 50k to token holders via USDC airdrop, with eligibility requiring organic launch on Bags, no bots or wash trading.
GSD 1.0 achieved 27.5k GitHub stars and 150k monthly downloads but was limited by Claude's codebase architecture, prompting a rebuild from scratch.
GSD 2.0 uses Piper, an open-source coding agent library that allows any model and custom extensions, solving the observability and autonomy constraints of Claude Code.
The autonomous agent ran for 9 hours straight building multiple projects, with new Slack and Discord notifications allowing users to check progress remotely.
GSD built an automated Stripe-to-token buyback system: card payments are processed through Stripe, profits accumulate in a wallet, then automatically swap to the token and burn it above thresholds.
Monetization strategy: charge non-crypto users $20–$200 monthly for a beautiful UI, with 50% of profits going to token holders through burns or treasury allocation.
Token-funded development: approximately $150–160k in SOL has funded full-time work, marketing via X posts, and model API costs; host noted the fee burn increases trading volumes and benefits other Bags apps.
GSD's core power comes from opinionated system prompts and custom tools, not just the underlying model; the agent must verify tasks rather than ask humans, with a browser tool and newly shipped Mac control via Xcode.
Planning redesigned from horizontal phases to vertical slices: each slice like 'authentication' touches database, API, and frontend, ensuring the agent can verify complete user-facing features end-to-end.
Automatic context clearing between slices prevents token bloat and context rot, enabling long-running autonomous sessions without degradation.
This summary was written from the recording's transcript by a language model. For the exact words, ask the MCG Live archive: every answer there is quoted word for word and linked to the second it was said. Market Bubble Search is independent and not affiliated with the show. Nothing here is advice.