Fere AI, a Singapore-based startup, has raised $1.3M in a seed round to build autonomous trading agents that adjust their own strategies based on real-time market feedback, according to Yellow.com and a GlobeNewswire announcement published April 23.
The Product
Fere’s core offering is a trading agent that runs continuously and refines its strategy over time without manual intervention from users. The company describes the system as “self-improving,” meaning the agent iterates on its approach using market feedback loops rather than static rulesets. The target market is retail crypto investors who currently lack access to the algorithmic tools available to institutional desks.
The company has not disclosed specific backers or the identity of lead investors in the round.
Context
At $1.3M, the raise sits at the smaller end of the AI agent funding spectrum, suggesting pre-product or very early-stage deployment. For comparison, NeoCognition raised $40M in April to build self-learning agents for enterprise, and Nava closed $8.3M for agent trust infrastructure earlier this month.
The broader category of AI-powered crypto trading agents has grown steadily through 2026, with multiple protocols launching agent frameworks on Solana and Ethereum. Retail-focused products have generally lagged institutional tools in sophistication, which is the gap Fere says it targets.
The Differentiation Question
“Self-improving” is a strong claim. Whether Fere’s agents genuinely learn and adapt in production, or simply optimize within pre-defined parameter boundaries, will determine whether this is a legitimate step toward autonomous financial agents or another backtesting wrapper with a better pitch. The company’s next milestone will be demonstrating live performance data against standard benchmarks.