Auto.fun

Eliza Labs had an open-source multi-agent framework and a large community, but launching an agent still required real engineering skill — and the standard bonding-curve launch model gave creators nothing after the initial spike while bots accumulated cheap supply.

Prototype a single flow where a non-technical creator describes an agent in plain language and the platform deploys both the agent and its paired token, with launch mechanics that favour real users over snipers and keep paying the creator over time.

• 01Built the no-code agent creation flow on top of the elizaOS framework, mapping plain-language intents to agent capabilities such as social posting, yield monitoring and scheduled execution.
• 02Implemented a 'fairer than fair' bonding curve in Anchor allowing teams to reserve up to 50% of supply pre-listing as an explicit anti-sniping measure.
• 03Designed liquidity-NFT fee routing so a defined share of ongoing swap fees returns to the original creator indefinitely rather than only to liquidity providers.
• 04Kept the entire codebase open source and reviewable, so users can inspect exactly what an agent does with their permissions and funds.
• 05Shipped a launch cohort tooling pack — deployment scripts, agent templates and monitoring — used by the first wave of projects on the platform.

• Working MVP with agent creation and token launch collapsed into one no-code flow.
• 15+ projects onboarded in the initial launch cohort across social, compute and payments agents.
• Creator earnings decoupled from the launch spike through perpetual fee routing.
• Fully open-source deployment, verifiable end to end by users.