As AI devours traditional web traffic, blockchain startups are reviving the dream of user-owned data and agent-powered micropayments.
The question of how to power the internet has been a stubborn source of debate since more or less the genesis of the information superhighway. (We mean “power” metaphorically speaking.) How to address this question has been far from monolithic, generating variations upon variations of solutions. On the one end, there are calls for open-source protocols and the generation of a cybernetic commons. On the other end, the more prominent one, there’s an enclosure of the commons: giving way to the revenue streams we know (and probably don’t love) today, like click-through ad income — not to mention the myriad platforms inhabiting the internet, now offering a wide menu of subscription structures and contract terms to best extract value. Pick your poison.
Between these two ends has long been a quieter camp advocating for micropayments as a way to keep the internet running and — along with the information and other value delivered through it — sustainably viable. Web browsers like Brave and Puma allow for sending small payments to websites, but, in general, the idea hasn’t really taken off.
However, the math isn’t really math-ing, especially as artificial intelligence continues to expand its hold over internet traffic. The WIRED AI Summit in September brought together a panel of media industry big wigs: Jim Bankoff, Cofounder, Chair, and CEO of Vox Media; Roger Lynch, CEO of Condé Nast; Mike Reed, CEO of the Gannett/USA TODAY Network; and Neil Vogel, CEO of People Inc. The execs described how internet traffic is tanking in the age of generative AI search results, leading to severe reductions in ad revenue, as consumers click through to websites via search engines far less frequently than they once did. Research from Pew suggests AI summaries may slash site traffic by up to 50%.
While these media empires are diversified — and are able to lean into events, commercial partnerships, and other parts of their businesses to continue to do well — it’s still “incredibly disruptive and incredibly difficult,” said Vogel of People. (Not to mention the foreboding prospects of mass media layoffs and bankruptcies for the smaller-scale publications that can’t weather this kind of storm, or lacked the operational breadth needed to run several revenue-generating verticals.) Vogel called out Google as being particularly combative, as its web crawler and AI crawler are the same tool, meaning that if a media company turns off Google’s access to its website, it’ll lose out on being listed on Google tout court — not just in AI summaries.
“The fundamental thing that Google is doing right now is using their market power to make sure we can’t take back leverage,” he said. Contrasting this practice with the slightly more diplomatic relationships media giants hold with OpenAI, Anthropic, and others shows Vogel that one maxim holds truest: “Garbage in, garbage out. You need high-quality content to make high-quality models.”
That adage has trickle-down implications, and AI startups know it. Leaning into the perceived synergies that may arise when tokenized protocols wed to agentic frameworks, outfits like Switzerland-based Permission have built AI agents that negotiate on users’ behalf to agree to see ads, sign up for market studies, and more, receiving micropayments through the $ASK token based on the amount of data a user shares with these various vendors. In essence, Permission helps you see and be rewarded for the work you already do to keep the internet functioning — perhaps to an extreme, albeit a mildly lucrative one.
Permission, which has raised $13 million since its founding in 2018, is far from the only player in this space, or attempting to operate a business model similar to it.
- On the flip side of Permission, for instance, there’s Masumi, which monetizes the efforts of agents, enabling agents to use programmable money to reward each other for different tasks within a multi-chain process (almost like a multi-partite micro-payments “handshake” among a series of bots).
- In the healthtech realm, there’s Cudis, Oura Ring-like tech that can connect to marketplaces that allow you to own your health data, and, potentially, monetize it.
- And then, further into the blockchain world, are collectives like ReflexDAO, which creates an ecosystem rewarding users for providing data — seemingly, in part, through the direct participation of researchers in areas that stand to benefit ecosystem participants.
The issue with some applications of the data-marketplace model, however, is that the model incentivizes data provision but doesn’t necessarily filter according to data quality, said Shady El Damaty, who builds protocols and technologies layered on privacy-preserving financial models. He added that, especially for these data-monetization protocols that depend on some sort of decentralization, “adding any sort of incentive on data also encourages farmers to Sybil attack with fake or simulated data.” A lack of data provenance will mean two-sided marketplaces are not trusted, and therefore may fail.
This doesn’t necessarily mean that there’s a massive whitespace for data-quality tooling. Solutions like Datafold, Amundsen, SodaGPT and others tend to gauge inputs along eight dimensions, flagging inconsistent, duplicative, and/or conspicuously timed and logged data. In the eyes of the Datafold team, the ‘Garbage In, Garbage Out’ adage “lacks nuance and ignores the fact that we have a lot of sophisticated tooling to manage ‘garbage.’”
Rather, the economic fundamentals involved in the decentralized, semi-self-directed provision of data are the major barrier, and constrain the use cases in which a tool like Permission can be viable and deployable. To boot, outsourcing data acquisition butts up against the range of trusted intermediaries in highly sensitive sectors — in healthcare, for instance, that means doctors’ offices, hospitals, and the like — to sign up relevant study participants. Rather than a question of tech, it’s one of trust, resilience against corruption, hedging against lawsuits, and plain old economics. For now, it seems, our data-driven micropayment handshakes are a currency of very few realms.
