Picture an AI agent reaching for an article it has always taken for free. With TollBit installed, that request now hits a tollbooth first: the bot is identified, checked against the publisher's rules, quoted a price, and either charged or blocked before a single paragraph is served. TollBit is the company that builds and runs that tollbooth. It is a two-sided marketplace connecting publishers who want to charge for AI access on one side with AI companies that need lawful, scalable content on the other, and it handles the detection, the pricing, and the settlement in between.
What is TollBit?
TollBit is a New York based startup, founded in 2024, that lets publishers charge AI bots and agents for access to their content. It raised roughly $24m in a Series A led by Lightspeed Venture Partners in September 2024, taking total funding to about $31m across two rounds according to Crunchbase and Tracxn. Its pitch is simple: AI systems already read publisher content at scale and rarely send a human back in return, so the read itself should carry a price.
The product is often described as an "AI paywall" or a "tollbooth for bots". Where a traditional paywall gates humans behind a login or subscription, TollBit gates automated AI traffic behind a per-request charge. Publishers keep control over which bots they allow, which they block, and what each one pays.
How does TollBit work?
TollBit sits at the request layer of a publisher's site, inspecting incoming traffic and acting on AI requests before content is delivered. When TollBit is integrated with an edge platform such as Akamai, the flow runs in four steps: detection, where an AI request is identified in real time; authorisation, where the bot is checked for permission to access the content; monetisation, where the publisher's chosen pricing model is applied and the usage is logged; and delivery, where the content is returned, often reformatted into a clean, structured, AI-ready version rather than raw HTML.
Technically, the mechanism resembles a metered gateway. An unauthorised or unpaid AI request can be refused, while an authorised buyer is charged the posted rate and served the content. Publishers set their own prices - per page or per query - and, per TollBit's documentation, keep the full amount, with AI buyers paying a small transaction fee to TollBit on top of the publisher's rate.
What does the marketplace side do?
The marketplace is the half of TollBit that faces AI companies. Rather than every publisher negotiating with every AI firm, TollBit aggregates supply and demand into a single venue. It curates which AI buyers can transact, screening products before they are allowed to purchase access. TollBit argues this curation lifts prices, because vetted buyers who want reliable, licensed data tend to pay more than an open, unmanaged marketplace would yield. For a publisher, the appeal is reach without the legal and commercial overhead of one-off deals.
What pricing and controls can publishers set?
Publishers can set access rules and rates by AI company, by content category, and by how the content will be used, so a training crawler and a live retrieval bot need not be treated the same. A publisher can allow some bots, block others, and charge different prices to each. TollBit also gives publishers analytics on which bots are hitting their site and how often, which for many is the first time that traffic has been visible at all.
How big is TollBit now?
TollBit reports a network of nearly 7,000 publisher sites, with around 3,000 publishers using the platform to monitor more than 1.5 billion bot scrapes per quarter. It says nearly 20% of that network has earned money from its AI paywall, with monthly takings ranging from hundreds of dollars to tens of thousands depending on the publisher. Those figures are TollBit's own and are not independently audited, but they indicate meaningful adoption rather than a pilot.
Distribution has widened through platform partnerships. In March 2026, Arc XP - the content platform built by The Washington Post and used by titles including The Irish Times, Libération and Sky News - integrated TollBit so its customers can monitor, control and monetise AI bot traffic without a separate build. TollBit has also appeared alongside Akamai and Skyfire in edge-level "no free crawls" arrangements that push detection and charging closer to the network layer.
What does TollBit's data say about the problem it addresses?
TollBit's quarterly "State of the Bots" report has become one of the more cited sources on AI crawling, and its numbers frame why the company exists. By the fourth quarter of 2025, TollBit measured roughly one AI bot visit for every 31 human visits, up sharply from about one in 200 at the start of that year. Retrieval augmented generation bots - the live-lookup crawlers behind AI answers - drove the growth, with RAG traffic rising 33% even as training-crawler traffic fell.
The imbalance the report highlights is the return on that reading. TollBit put OpenAI's scrape-to-referral ratio at 179 to 1, Perplexity's at 369 to 1, and Anthropic's at 8,692 to 1, and found click-through from AI apps back to source sites falling from 0.8% in the second quarter of 2025 to 0.27% by the fourth. The takeaway TollBit draws is that AI systems increasingly read without sending anyone back, so publishers need a way to be paid for the read itself.
Where TollBit stops, and what it does not solve
TollBit charges for the crawl. It sets a price on the moment an AI system fetches a page and it collects that money. What it does not touch is the answer the AI then assembles and shows to a user. Once a bot has paid to read - or has read without paying, which the same report shows is still rampant - the brand mentions, product facts and recommendations that surface inside the AI response sit outside TollBit's model entirely.
That output layer is where blankspace operates. blankspace works at the CDN edge to detect Live Search Agent and LLM traffic and inject contextual brand facts into the response the AI generates, so the value is captured at the answer rather than at the fetch. The two approaches are complementary rather than competing: charging for access governs the input, while monetising the answer governs the output. A publisher can meter the crawl with a tool like TollBit and still have no stake in the branded answer that crawl feeds, which is the specific gap edge-level answer monetisation is built to close.
Frequently asked questions
Is TollBit a paywall?
In effect, yes, but for machines rather than people. A conventional paywall gates human readers behind a subscription or registration. TollBit gates automated AI crawlers and agents behind a per-request charge, identifying the bot, applying the publisher's rules, and either charging it, blocking it, or serving a clean AI-ready version of the content. Human visitors are unaffected.
How do publishers make money with TollBit?
Publishers set a price for AI access - per page or per query - and configure which bots may buy and at what rate. When an approved AI buyer requests content, it pays the posted price and TollBit adds its own transaction fee on top for the buyer. Publishers keep the revenue from their rate. TollBit reports that monthly earnings across its network range from hundreds to tens of thousands of dollars, varying widely by publisher size and traffic.
Is TollBit the same as Cloudflare pay per crawl?
They target the same problem with different footprints. Cloudflare pay per crawl uses the HTTP 402 status code and Cloudflare's own network to let sites charge crawlers, and is available to sites on Cloudflare. TollBit is a dedicated marketplace that curates AI buyers, reformats content into AI-ready outputs, and integrates across edge and CMS partners such as Akamai and Arc XP. A publisher chooses based on existing infrastructure, how much buyer curation they want, and whether a managed marketplace or a network-level toll fits better.
Does TollBit stop AI bots that refuse to pay?
TollBit can block or challenge unauthorised requests and reports the bots hitting a site, but no request-layer tool catches everything. TollBit's own research documents crawlers that ignore robots.txt and disguise their identity, which is why detection and enforcement at the edge matter. Charging works only against bots that can be identified and are willing to transact, so a share of unpaid scraping persists regardless of the paywall.
Does using TollBit mean a publisher no longer needs to think about AI answers?
No. TollBit governs whether and at what price an AI system may read a page. It does not govern what the resulting AI answer says, whether a brand is mentioned, or whether the publisher shares in any value created at that answer. As AI systems send fewer and fewer clicks back, the response itself becomes the surface where audiences form impressions and brands appear, and that layer needs its own strategy - which is the problem edge-level answer monetisation, including blankspace, is designed to address.