Parallel Launches Marketplace to Pay Publishers for AI Agent Work
Parallel founder and former Twitter CEO Parag Agrawal argues that AI agents are breaking the web’s existing content economics by using publisher and creator material to perform valuable work without tying compensation to that value. In a Bloomberg Technology interview, Agrawal said Parallel’s new Index marketplace is meant to pay publishers, data providers, and independent creators according to their content’s measured contribution to an agent’s completed task, rather than through ads, subscriptions, citations, or flat licensing deals.
Parallel wants content owners paid for what agents actually accomplish
Parag Agrawal says Parallel’s new Index marketplace is built around a break in the web’s economics: AI agents may use web content far more heavily than humans, while the systems that supported publishing were designed around human attention.
Parallel started, Agrawal said, from the assumption that “agents will use all of the content on the web thousand x more than humans ever have.” In that world, ads and subscriptions no longer map cleanly to how content creates value. Index is Parallel’s attempt to create a market between agents and the entities deploying them, and the publishers, creators, and data providers whose material helps those agents complete tasks.
That differs from the flat-fee licensing deals Agrawal described elsewhere in the market, typically between a large AI lab or hyperscaler and a large publishing house. His objection is not that those deals pay nothing. It is that fixed payments do not let content owners share in the upside as agents begin doing more valuable work.
When they produce value, content owners don't get to participate in that economy of the value being created.
Agrawal’s examples were not consumer search summaries. He pointed to customers building AI scientists, AI lawyers, and finance-focused agents — systems meant to perform “real knowledge work in the enterprise.” Parallel’s argument is that a source should be worth more when it supports a valuable task, and less when it is merely present but not decisive.
Parallel lists Sequoia, Kleiner Perkins, Index Ventures, Khosla Ventures, Spark Capital, and Abstract Ventures among its backers. Index’s public product pitch is addressed directly to publishers and data owners: “See what your content is worth to AI agents,” with a domain search bar inviting checks for sites such as arxiv.org, nyc.gov, and propublica.org.
The pricing model is based on marginal contribution, not just access or citation
The central pricing question is who sets value when an agent uses a source: the agent side, the content owner, or some measured contribution between them. Agrawal pointed to a model Parallel says it has built to estimate “marginal contribution.” The concept behind it, he said, comes from Shapley values, a game-theoretic method for dividing the value created by a collaboration.
The practical problem is harder than pricing content access, crawls, or citations. The monetary number has to be pegged to the value of the work an AI agent performs, while accounting for what a given source contributed to that result.
Agrawal described Shapley values as a way to answer a problem in positive-sum collaboration: when multiple contributors help create something whose total value is larger than the sum of the individual parts, how should that value be divided so everyone remains incentivized to participate? Applied to content used by agents, he said the model produces three properties Parallel wants: higher-quality content gets paid more; content used in higher-value work gets paid more; and content owners grow economically as agents perform more valuable work.
As agents do more work and create more value, content owners grow alongside them, which we believe allows content creation on the web to be sustainable.
Index’s dashboard illustrates the kind of measurements Parallel is presenting to publishers. A displayed nyc.gov example describes the domain as “difficult to substitute in the answers it appears in,” with “marginal contribution” running ahead of demand. The dashboard scores domains across impressions, citations, value, and uniqueness.
| Metric | What the displayed dashboard said it measures | nyc.gov score shown |
|---|---|---|
| Impressions | How often agents surface the content while researching, even if it does not appear in the final answer | 77/100 |
| Citations | How often the content gets cited after being considered | 72/100 |
| Value | How valuable the tasks citing the content are | 74/100 |
| Uniqueness | How the domain ranks against substitutes agents consider for similar tasks | 78/100 |
The same interface compared citation share among domains: nih.gov captured 81.7% of citation attention among the domains shown, while nyc.gov captured 13.9%; nasa.gov, sf.gov, and loc.gov had smaller shares. The dashboard also offered tabs for “Share of Citations,” “Share of Value,” and “Share of Unique Value,” underscoring Parallel’s argument that citation alone is not the full accounting unit.
Parallel says the market should include small creators and smaller agents
Caroline Hyde pressed on the distribution problem: valuable content is not confined to large publishers or prominent individual writers. A smaller independent publisher could have material that is unusually useful to an agent, but the obvious participants in licensing markets tend to be large media companies or better-known creators.
Agrawal said Parallel is launching Index with partners across several categories. The named launch partners shown on screen include Fortune, RocketReach, The Atlantic, Enigma, Fiscal.ai, and ZoomInfo. Agrawal’s own list included premium news publishers such as The Atlantic and Fortune; “raw factual” content from PR Newswire; business intelligence and data providers including PitchBook, Tracxn, and ZoomInfo; and independent creators including Alex Heath and Azeem Azhar.
The broader goal, he said, is for content owners “big and small” to participate, and for all agents — not only those built by large providers — to access the content. Parallel’s pitch is therefore not limited to compensating large publishers already positioned to negotiate licensing deals. Agrawal described a market in which smaller content owners can participate and agents beyond the largest providers can get access to “amazing content.”
Index’s website frames the shift in broader web terms: “The web was built for humans. Our clicks and our eyeballs shaped how information got discovered and monetized.” The same text says the web’s primary user is changing, with AI agents “displacing the attention that built the content economy, without replacing the revenue that sustained it.”
That is the core tension Agrawal returned to throughout: if AI systems consume and operationalize content at scale, but the revenue mechanisms remain tied to human attention, creators and publishers may lose the economic link that sustained content creation on the web. Index is Parallel’s attempt to build a payment model around measured contribution to agent work.
The open-web argument is access plus compensation
Ed Ludlow briefly turned to X, the platform formerly known as Twitter. Agrawal said he still uses it and still calls it Twitter, and described the product built during his decade there as enduring in value. The connection to Parallel was thematic: he said Twitter reflected “content access and democratization out in the open” rather than in permissioned groups.
That same preference shapes how Agrawal described Parallel. Index is not presented as a way to wall content off from AI systems. It is presented as a way to keep content broadly usable while changing how content owners are paid for the value agents derive from it — “everyone’s agents” having access, rather than only a few.