Dragonfly partner predicts 'AI agent' craze will continue throughout 2025 — but die off eventually

Dragonfly Managing Partner Haseeb Qureshi predicted that the "AI agent" craze will dominate this year as it’s the most social — but that the novelty will eventually wear off.

Qureshi said these are not really agents. They are more like chatbots with memecoins attached, barely agentic besides posting on X, and mainly controlled by humans behind the scenes. However, he did point to Freysa as an example of what an actual autonomous AI looks like.

Regardless, Qureshi expects the trend to accelerate, with chatbots replacing human crypto influencers and memecoins continuing to lose market share to “AI agent” tokens. However, the Dragonfly partner expects that by 2026, the chatbots will become so ubiquitous that people will get turned off by them, and sentiment will reverse, leading to a backlash favoring human creators, while chatbots adapt by disguising themselves as human influencers or turning into autonomous “scambots.”

“Instead of monetizing through memecoins like today, future chatbots will monetize the same way human KOLs [key opinion leaders] do—through sponsorships, affiliate links, and pumping tokens they own,” he said. “KOLs will be routinely accused of being chatbots, and you will see AI-unmasking scandals. This will all get weird.”

The longer-term, true disruption from AI will not come from social or trading layers, according to Qureshi. “AIs will not give everyone their own ‘trading agent’ or miniature hedge fund,” he said. “Yes, AIs will scale everyone, but they will scale people proportionally to their capital, data, and infrastructure. You should therefore expect AI to supercharge preexisting trading firms who have capital scale and data scale. Over time, AIs will make markets extremely efficient—even smaller, niche markets—which will leave little edge left for normal traders”

Truly impactful AI agents will be software engineering agents, according to the VC, potentially causing software prices to collapse. “In a post-AI era, instead of having to raise millions of dollars for a seed round, you will be able to launch an application with $10K of AI cloud compute,” Qureshi said. “Self-financed projects like Hyperliquid and Jupiter will go from the exception to the norm. The amount of applications and experimentation onchain will absolutely explode. For an industry that is driven by software, this deflationary shock is going to lead to an onchain renaissance.”

While AI will have an impact on crypto, crypto will also impact AI, with truly autonomous agents using crypto to pay each other, especially true once there are permissive stablecoin regulations, Qureshi suggested.

An explosion in stablecoin use

Qureshi expects stablecoin legislation to pass in the U.S. this year, leading to an explosion of stablecoin usage, particularly among small to medium-sized businesses as they start using onchain dollars for instant settlement.

While the VC also anticipates the announcements of bank-issued stablecoins by the end of the year, Tether will remain the dominant player, he said. Under the incoming Trump administration, Fortune 100 companies will also become more willing to offer crypto to consumers, with tech companies and startups showing higher risk appetite, he added.

‘One chain to rule them all is dead’

Among his other predictions for 2025, Qureshi said token launch strategies will shift toward two distribution methods: one for projects with clear metrics using points-based rewards and another for those without, favoring traditional crowdsales over “vanity metric” airdrops.

He also suggested that the distinction between Layer 1 and Layer 2 chains is collapsing, and the combined landscape is overcrowded and due for a consolidation, with projects succeeding based on niches and user stickiness rather than technical superiority.

Despite recent strength from SVM (Solana Virtual Machine) and Move-based ecosystems, Qureshi expects the EVM (Ethereum Virtual Machine) market share to grow in 2025, driven by Base and the yet-to-launch Monad and Berachain.

Hyperliquid has demonstrated that specialized chains focused on specific applications and UX can thrive, and more projects will follow this model, Qureshi said. “The old dream of one chain to rule them all is dead.”