Nansen co-founder and CEO Alex Svanevik said publicly on July 28 that AI agents will surpass human traders within two years, a prediction he is backing with a full product overhaul at the company he leads. The analytics firm, founded in 2019 and long known for its wallet-labeling database, now describes itself as an agentic trading platform. That repositioning carries structural weight: Nansen moves from a neutral data provider to an active participant in on-chain trading execution, a model closer to a vertically integrated brokerage than a research terminal. Svanevik is therefore forecasting an outcome that directly benefits Nansen's new commercial positioning, a context readers should weigh when evaluating his two-year timeline.

The platform pivot began on January 21, 2026, when Nansen launched what it calls "vibe trading," a conversational interface that lets users describe a trade in plain language and have an AI agent identify opportunities, surface supporting on-chain signals, and execute the order. Execution routes through Jupiter on Solana, OKX, and the cross-chain protocol LI.FI. Nansen has since embedded Hyperliquid perpetuals trading directly into the same interface, letting a user spot whale wallet activity and open a leveraged position without switching apps.

"Traders want one place where they can get exposure to different assets, see who's winning onchain, and act on it," Svanevik said in remarks tied to the Hyperliquid integration.

The technical edge Nansen claims rests on its database of more than 500 million labeled wallet addresses. General-purpose large language models, when operating without access to proprietary on-chain data, lack visibility into this layer. Nansen addresses that gap by routing leading models, including GPT and Claude, through its proprietary data rather than building a competing model from scratch. "We're giving the best models eyes on-chain," Svanevik said. According to Nansen's own quality evaluations, the provenance and independence of which the company has not publicly specified, the platform's Expert mode scores 85% on a quality evaluation, compared with roughly 20% for models running the same tasks without the on-chain data layer.

Svanevik is not arguing for immediate full autonomy. He frames adoption in three stages: first, an agent that recommends trades for human approval; second, an agent that surfaces AI-suggested trade opportunities; third, fully autonomous execution. Skipping ahead to stage three today would be, in his words, "a really bad idea," comparing it to sitting into a Tesla for the first time ever and going straight to the back seat and letting it drive. He also disclosed that Nansen itself runs 75 Claude agents internally across company operations.

The broader market context underlines the scale of the shift underway. On-chain perpetual trading volume across all platforms exceeded two trillion dollars in Q1 2026. Hyperliquid alone accounted for 625 billion dollars of that, cementing its position as the leading decentralized derivatives venue. Approximately 30% of Hyperliquid traders also trade equities, commodities, and index-linked perpetuals, signalling growing crossover into broader asset classes. The AI crypto sector's total market capitalization grew from roughly nine billion dollars in early 2025 to somewhere between 22.6 and 27 billion dollars by May 2026. Across blockchain networks, autonomous AI agent deployments surpassed 20,000 by February 2026, a 300% increase from Q4 2025. About 41% of crypto hedge funds and institutional trading firms now report actively using or testing on-chain AI agents.

The implications extend well beyond North America and Europe. India ranked first on the 2025 Chainalysis Global Crypto Adoption Index, and Pakistan placed in the top five globally. APAC crypto transaction volume nearly doubled to 2.36 trillion dollars. For retail traders in South Asia, an agentic interface that replaces complex charting dashboards with a plain-language prompt could substantially lower the barrier to entry. The model maps naturally to mobile-first usage patterns common across India, Pakistan, and Bangladesh, markets where DeFi participation is already disproportionately high relative to global averages. India's Digital Personal Data Protection Act also introduces concrete data sovereignty considerations for South Asian users of a platform whose core competitive asset is a 500 million-address behavioural database, a regulatory dimension that remains unresolved. In Africa, particularly Nigeria, South Africa, and Kenya, AI trading tools are already gaining traction as local brokers respond to demand. For users in high-inflation, FX-restricted environments, an agent capable of autonomously managing stablecoin rotations or DeFi yield positions without constant manual oversight addresses a real practical need.

Risks are real and currently under-addressed. A 2026 security incident documented by KuCoin research involving autonomous AI trading agents resulted in more than 45 million dollars in breached funds, with attack vectors including memory poisoning, indirect prompt injection, and weak context handling. Retail traders in markets with limited consumer protection frameworks carry the most exposure if an agent executes an erroneous or manipulated trade. Regulatory clarity is absent in most emerging markets; neither South Asian nor African regulators have established frameworks for how autonomous on-chain trading entities should be classified or held accountable.

Svanevik's longer forecast calls for conversational AI to replace dashboards as the dominant investor interface by 2030, and for ETFs and traditional finance wrappers to fade out by 2040 to 2050 as retail investors trade directly on-chain. "Trading in 2030 is going to look very different from trading up to now," he said. Whether the two-year timeline for AI to overtake human traders holds will depend heavily on how quickly the trust and security problems get solved, not just how good the models get.